Word word.... I might have stuck with 10.04 though... IDK kernel upgrade would be nice even if I'm not sure about the new feature set helping you... updated Eucalyptus and now Open Stack MIGHT be nice, but that would require a ton of changes and stuff....
IDK what kernel you were rocking the lamps on though, and how much traffic you were getting... I know the big kernel lock removal was a big deal to a lot of people, but I'd think that would be more for server farms and shit verses a LAMP stack...
Let us know how the migration goes.
If I may probe further because I'm curious? I assume you're rocking the server version of Ubuntu? Any reason you and your clients chose Ubuntu for the LAMP verses many other distros people tend to think of for server needs? (I have my own random thoughts on this, but won't bore others here).
Sure thing. Kernel upgrade and other things were part of it. The current kernel 2.6.32-31 on a majority of the servers I work on. I figured it was time for me to update since I can still access the older kernels through a VM for testing if need be before production sites get touched.
As far as the LAMP stack, yes, I've got LAMP running on 10.04LTS servers. In terms of choosing it, I'm very comfortable with Ubuntu and hardening servers in a debian based environment. A lot of friends/consultants use Fedora or CentOS which I get is reliable and strong, but I haven't had any issues with my Ubuntu servers.
I'd agree he doesn't mention details though I assume that's what he thinks the actual summit sessions are for. Still would have been nice to hear some details though.
I can't fucking quote that enough. It's one of the areas that makes me want to pull my hair out as a relatively new user in the linux space that considers himself casual which makes me IDK have a slighly better read on other casual users. Now I like the idea of having an open source alternative for everything, but gosh damn go on some linux forums and they'll bend you over for even thinking about using proprietary software even if the open source variant just isn't up to par. They'd rather you use the open source version just because even if it sucks.
Again I'd like to hear more details, and maybe they are hashing it out. Yet while over the past few years of getting into Linux and Ubuntu I've never been a big Shuttleworth fan he struck me as someone who "got it" where as(hate on me if you will) a lot of the hardcore GNU/Linux nerds don't in terms of the broader population.
Well, just look at something like Truecrypt. Open source and gpl like license. You can encrypt your whole windows hard drive and do pre-boot authentication.
But it's not available on any of the repositories for distros because the license is not free enough. Instead you have to walk around, play with lvm, cryptsetup, hashalot, modify /etc/crypttab and /etc/fstab in order to have a simple pre-boot authentication so your data is safe.
The "true open source" alternative is much much worse, but people don't give a fuck. They stick to their ideas not caring a bit about it. Meanwhile I have to go troughs 800k steps in order to do something that is one click on windows, just because is not true open source. Fuck that.
Oh, and the problem with Ubuntu just doesn't come from casual or hardcore. Its hte mismanagement that seems to be behind it and the poor community communication at the end of the day. Not working with upstream also gave them some (deserved IMO) flak.
But in the end what he says is nothing new.
People want easy things to manage.
People want candy desktops.
People don't care about open source.
And that is where ubuntu has been heading since day 1.
BTW, if anyone is interested in setting their debian based distro partition encrypted with pre-boot authentication let me know. Its a pain in the ass to do and there is little and outdated info out there so I'll gladly help if anyone needs it.
Question for those Linux freaks out there... anyone have any experience with Zenoss? i'm getting it installed on a 10.04 box and just wondering what if any thoughts were on it? Basically we have no monitoring of any kind at work and I feel obligated to learn how this works until we move to the commercial version or Solarwinds. Anyways, just curious.
Well, just look at something like Truecrypt. Open source and gpl like license. You can encrypt your whole windows hard drive and do pre-boot authentication.
But it's not available on any of the repositories for distros because the license is not free enough. Instead you have to walk around, play with lvm, cryptsetup, hashalot, modify /etc/crypttab and /etc/fstab in order to have a simple pre-boot authentication so your data is safe.
sudo zypper info realcrypt
Loading repository data...
Reading installed packages...
Information for package realcrypt:
Repository: http://packman.iu-bremen.de/suse/11.3
Name: realcrypt
Version: 7.0a-3.4
Arch: x86_64
Vendor: http://packman.links2linux.de
Installed: No
Status: not installed
Installed Size: 2.9 MiB
Summary: Free Open-Source Disk Encryption Software
Description:
Based on TrueCrypt, freely available at http://www.truecrypt.org/.
RealCrypt is mainly just a rebrand to allow for modifications to take place,
functionality remains all the same.
sudo yum info truecrypt
Name : truecrypt
Arch : x86_64
Version : 6.2a
Release : 2.el5.rf
Size : 1.8 M
Repo : rpmforge
Summary : Free open-source disk encryption software
URL : http://www.truecrypt.org/
License : TrueCrypt License 2.7
Description: Manages encrypted TrueCrypt volumes, which can be mapped as virtual block
: devices and used as any other standard block device. All data being read
: from a mapped TrueCrypt volume is transparently decrypted and all data
: being written to it is transparently encrypted.
Different distros have different philosophies, different levels of trust, etc., in their respective repositories. Ubuntu in particular likes to be different and shouldn't be treated as a general synonym for Linux. Most prominent Linux distros seem to support TrueCrypt in either their main repositories or others that can be readily added. In the case of SuSE, they probably had to make a minor source code change to get it to compile for their distribution, and (I'm told) the TrueCrypt license requires that you change the name. FreeBSD doesn't have it, but that's largely because of an OS compatibility (ie, it's not actually Linux, so there are technical issues). PCLinuxOS seems to have packages for it, but I don't know if it's in any repository. Mandriva lists the issue as resolved. I don't know about Slackware -- there's a "slackbuild" for it on some dude's website, but I don't actually know what that means, having never used Slack.
Is the above url the last we've heard of the situation? At least some of the complaints about the license seem pretty valid. The "NOTHING IN THIS LICENSE SHALL IMPLY ... A PROMISE NOT TO SUE FOR COPYRIGHT INFRINGEMENT" was certainly pretty obviously problematic, though they have since changed the wording of at least that part in a way that it is incredibly redundant but much more commonsensical (eg, instead of saying "we can sue you for using our product regardless of whether or not you comply with this agreement", they are now saying "we can sue you for using our product if you do not comply with this agreement"). Understand that Apple probably would not allow something like that in their "App Store" repository, and Microsoft also would head to the hills if the wording were as such. This isn't an "open source" thing. It's a "protecting your legal ass" thing.
Anyway, the complaints may be totally out of date, but I'm slightly too lazy to go over that url point by point to see if all the issues were addressed. But considering that the vast majority of the Linux packaging world seems to not have a problem with it, it is not impossible to infer that this particular situation was resolved
(unless you use one specific particular brand of Linux, assuming that it represents the rest of them)
Different distros have different philosophies, different levels of trust, etc., in their respective repositories. Ubuntu in particular likes to be different and shouldn't be treated as a general synonym for Linux. Most prominent Linux distros seem to support TrueCrypt in either their main repositories or others that can be readily added. In the case of SuSE, they probably had to make a minor source code change to get it to compile for their distribution, and (I'm told) the TrueCrypt license requires that you change the name. FreeBSD doesn't have it, but that's largely because of an OS compatibility (ie, it's not actually Linux, so there are technical issues). PCLinuxOS seems to have packages for it, but I don't know if it's in any repository. Mandriva lists the issue as resolved. I don't know about Slackware -- there's a "slackbuild" for it on some dude's website, but I don't actually know what that means, having never used Slack.
Is the above url the last we've heard of the situation? At least some of the complaints about the license seem pretty valid. The "NOTHING IN THIS LICENSE SHALL IMPLY ... A PROMISE NOT TO SUE FOR COPYRIGHT INFRINGEMENT" was certainly pretty obviously problematic, though they have since changed the wording of at least that part in a way that it is incredibly redundant but much more commonsensical (eg, instead of saying "we can sue you for using our product regardless of whether or not you comply with this agreement", they are now saying "we can sue you for using our product if you do not comply with this agreement"). Understand that Apple probably would not allow something like that in their "App Store" repository, and Microsoft also would head to the hills if the wording were as such. This isn't an "open source" thing. It's a "protecting your legal ass" thing.
Anyway, the complaints may be totally out of date, but I'm slightly too lazy to go over that url point by point to see if all the issues were addressed. But considering that the vast majority of the Linux packaging world seems to not have a problem with it, it is not impossible to infer that this particular situation was resolved
(unless you use one specific particular brand of Linux, assuming that it represents the rest of them)
I think I explained myself wrongly. What I meant its that because of license shenanigans we don't have an easy pre-boot authentication method available where there is a program that can do it.
The basis of it is that as it's not on the main repositories because of that, it can't be fully implemented on any distros.
How good will it be to have in the installer an option to encrypt the whole partition instead of playing with cryptsetup and lvm (Which as far as I know still needs you to manually add entries to /etc/crypttab) as it's the only method to do pre-boot authentication?
In the end my point is, that yes open source is incredibly good, but isn't the solution to everything. We got Debian for a true open source distro, the rest of them should be more loose with license and shit which in the end, the end user doesn't give a fuck about.
It reminds me of the dvdrtools discussion which keeps appearing from time to time. End users just want to burn dvds, they don't care about if the program its open or not, or the license says X instead of Y. And that is where linux as a whole should be heading towards in order to expand.
IIRC all the repos you posted are extra or have to be manually added(like rpmforge) except arch. Which is expected because arch seems to be the most open distro out there
In the end my point is, that yes open source is incredibly good, but isn't the solution to everything. We got Debian for a true open source distro, the rest of them should be more loose with license and shit which in the end, the end user doesn't give a fuck about.
Depends on which group of end users you're talking about. Plenty of people (including myself) care about the licenses of installed software. As for truecrypt I really suggest you go and read it's license. It's extremely restrictive and basically requiring you advertise if you include it which is why you won't find it in any installers. If I did manage a distro I wouldn't include it in the installer either.
Phrase "Based on TrueCrypt, freely available at http://www.truecrypt.org/" must be displayed by Your Product (if technically feasible) and contained in its documentation. Alternatively, if This Product or its portion You included in Your Product constitutes only a minor portion of Your Product, phrase "Portions of this product are based in part on TrueCrypt, freely available at http://www.truecrypt.org/" may be displayed instead. In each of the cases mentioned above in this paragraph, "http://www.truecrypt.org/" must be a hyperlink (if technically feasible) pointing to http://www.truecrypt.org/ and You may freely choose the location within the user interface (if there is any) of Your Product (e.g., an "About" window, etc.) and the way in which Your Product will display the respective phrase.
Your Product (and any associated materials, e.g., the documentation, the content of the official web site of Your Product, etc.) must not present any Internet address containing the domain name truecrypt.org (or any domain name that forwards to the domain name truecrypt.org) in a manner that might suggest that it is where information about Your Product may be obtained or where bugs found in Your Product may be reported or where support for Your Product may be available or otherwise attempt to indicate that the domain name truecrypt.org is associated with Your Product
Alright Linux-GAF. You guys and gals seem extremely knowledgeable. I have a bunch of questions mainly server related just about general server stuff cause I'm curious, wondering, have some questions in my mind, and would like to learn.
Would you guys hate me if I posted like a 10 question list? I don't want to use and abuse this thread so I figured I'd ask this before hand! Thanks in advance!
Alright Linux-GAF. You guys and gals seem extremely knowledgeable. I have a bunch of questions mainly server related just about general server stuff cause I'm curious, wondering, have some questions in my mind, and would like to learn.
Would you guys hate me if I posted like a 10 question list? I don't want to use and abuse this thread so I figured I'd ask this before hand! Thanks in advance!
Depends on which group of end users you're talking about. Plenty of people (including myself) care about the licenses of installed software. As for truecrypt I really suggest you go and read it's license. It's extremely restrictive and basically requiring you advertise if you include it which is why you won't find it in any installers. If I did manage a distro I wouldn't include it in the installer either.
Section III is full of stuff like that. It is not a free software license.
See, that is where I don't agree. You got the source code and you are free to modify and distribute it and only requires you to mention that is based on their product. A good encryption software is a good trade for a simple notice.
Bare in mind that Im no lawyer and english is my second language so I maybe be misinterpreting the license but I don't see anything wrong with it. In fact is pretty much GPL like in terms of redistributing and derivative works.
And when I talk about users I mean the casual users. My mother. My girlfriend. My friend, etc...
They don't care about licenses, they only want to use good software. If its pay software they will either pay or the more common route, they will pirate it. They will crawl shitty download sites, wait days for it to download, manage shitty keygens/cracks and all because its good software.
We have to ask ourselves as linux users, what do we want? Do we stay closed in respect to licenses not allowing anything to go into distros in order to maintain the "purity" of linux like debian with linux-firmware? Or do we embrace the possibility of using proprietary software which collide with the open source movement in order to gain userbase?
I always have been fond of the second one. The more users we got, the better our software will be, the more developers we attract, the more developers see that open source works and hopefully contribute to it.
Of course, then there is the other view of it. If there is propietary software that fills a void in linux, there is no reason to make an open source alternative because the proprietary one is what users use. No idea how to tackle that to be honest.
Oh, and I care about licenses. The only software I use on linux which is not gpl is skype (and it can die in a fire, fucking piece of shit). But I know the shortcomings of it, like not being able to play any game because of the experimental status of nouveau 3D. I am willing to put up with that and help the developers make a decent alternative to the nvidia driver. But as I said, casual users just want to play a game wich runs on linux, they don't care about the license of their display drivers. Its all about perspective and future
Fake edit: Im not being aggressive or anything, am I? Don't want to appear like that, as I said English is not my native language so excuse me if something sounds bad, not my intention
Alright Linux-GAF. You guys and gals seem extremely knowledgeable. I have a bunch of questions mainly server related just about general server stuff cause I'm curious, wondering, have some questions in my mind, and would like to learn.
Would you guys hate me if I posted like a 10 question list? I don't want to use and abuse this thread so I figured I'd ask this before hand! Thanks in advance!
And when I talk about users I mean the casual users. My mother. My girlfriend. My friend, etc...
They don't care about licenses, they only want to use good software. If its pay software they will either pay or the more common route, they will pirate it. They will crawl shitty download sites, wait days for it to download, manage shitty keygens/cracks and all because its good software.
Here's my third try. I keep making my responses too big and getting into the details of licenses, which you really aren't complaining about, per se. Here's my newest response:
In Linux-based operating systems, you often have a non-commercial version and a commercial version. The commercial version has proprietary drivers and additional programs that have more interesting licenses. The non-commercial version does not have enough legal or monetary protection to include these things.
In Windows- and Mac- based operating systems, you have a commercial version. The commercial version may or may not have proprietary drivers and additional programs that have more interesting licenses. There is no non-commercial version.
So if an end user wants to transition from a commercial version of one operating system (for example, Windows), but they want to retain the advantages of having a commercial version, then they merely need to use the commercial version of the other operating system (for example, Linux).
Regarding TrueCrypt in particular:
I'd need to be a lawyer to properly process the legal issues; however, I note that (to my recollection), it was never used by even the commercial versions. This could be for several reasons, and here's one: In SuSE, I believe that the installer includes or has included the option to encrypt volumes (this may even have been in opensuse, actually). Now, my research tells me that this does not include the /boot partition, which I think means that it does not meet your needs, but it may have been considered "good enough" for the vast majority of users, and there might not have been any demand to otherwise include it. I just don't know.
I have an honest question to which I seriously do not know the answer: Do the installers for latest consumer versions of Windows and OS X include the option to encrypt the hard drive upon installation?
Okay so lets preface this a little bit. Sorry I'm such a newb in this area. My experience with severs is basically well 0. I know CentOS seems to be the go to linux server, and is basically the free offspring of RHEL which is the other big nix server distro is you want to pay. Beyond that if your from the other camp Debian seems to be where it's at since like RHEL it's the slow , stable, not quick to change, but just works of all the well debian based distros.
Anyways as I stated a week or two ago I got a free sata HDD from an old comp, and since I dual boot I decided to make that linux only. Since I had a ton of space and only had Ubuntu 11.04 desktop on there I got bored late last Saturday night, and decided I'd try out Ubuntu Server. Sure it's not one of the aforementioned distros. Yet it's a new version that just hit, and I'm sort of an Ubuntu guy.
I say this because some of this is general questions, and some of this might just be how Ubuntu handles shit. WTF do I know honestly.... So here it goes....
1) How can you tell what programs you can install?
So I ran my install, partitioned my drive, and everything went fine. Then it allowed you to pick from some big services people normally run or just go ahead. I decided fuck lets install the LAMP stack, and that worked fine. Set up my user name password etc... etc... and got to the terminal screen.
I was like well wtf now? This was your idea after all (talking to myself in my head LOL)! I decided lets try and install a few things. So I ran the normal sudo apt-get install myphpadmin and that ran fine. Did the same thing for wordpress. Ditto for Emacs.
I assume these all worked because they were already in the repository. Yet at the command line interface how the fuck are you suppose to know? In desktop say Ubuntu 11.04 or the Fedora 15 beta with Gnome 3 you have your package manager. You searched and it came up. If not you hit the interwebs for either a .deb/rpm, tar, or say a PPA for Ubuntu. There seemed to be know way to visually see the package manager so I feel like I just got lucky that the shit I tried to install was there.
2) How do you manage users?
The install had me create my user name and password to login with, but I was sort of confused. Does that initial account I created mean it's the master say admin account? What about creating other accounts? How would you manage what users have access to and such (would you call this group policy?)?
3) Would the server distros really only be good for enterprise solutions?
Are the server distros just good for hosting large databases, and file storage and the like for large corporations? You know like running the London Stock Exchange? What about personal usage such as a media server, a backup server, or hosting your own site?
4) It appeared the top answer aka default for what to do with updates was manually manage on your own? Is this really the case?
The 2nd option was for automatically installing security updates and the like. Why wouldn't I want it to automatically update or at least auto notify me (I think this was the 3rd option) about updates? Just afraid updates might break shit? Wouldn't you want the security updates installed ASAP?
5) Is it true you can't host your own server (say wordpress blog or ftp etc...) because the ISP will block you even if you can get it setup?
Like say you wanted to run a small time wordpress blog. Is it true you can't get the connection out even if you can get wordpress installed on a lamp stack? They just port block or something?
6) Other than the GUI and what comes preinstalled depending upon the distro is there really any difference between a desktop and server distro?
I mean is the only difference between the two the fact that one boots directly to a terminal without the GUI overhead? Obviously there are differences in what's preinstalled, but you could install/uninstall to get the same packaged and stuff on both for them to be the same right? I mean the kernel itself is the same for both flavors of linux right?
IDK I feel like I have more questions, but I'll start with those for now. Time to go for my nightly walk to get some exercise! Thanks in advance!
I say this because some of this is general questions, and some of this might just be how Ubuntu handles shit. WTF do I know honestly.... So here it goes....
1) How can you tell what programs you can install?
2) How do you manage users?
3) Would the server distros really only be good for enterprise solutions?
4) It appeared the top answer aka default for what to do with updates was manually manage on your own? Is this really the case?
5) Is it true you can't host your own server (say wordpress blog or ftp etc...) because the ISP will block you even if you can get it setup?
6) Other than the GUI and what comes preinstalled depending upon the distro is there really any difference between a desktop and server distro?
IDK I feel like I have more questions, but I'll start with those for now. Time to go for my nightly walk to get some exercise! Thanks in advance!
1) In CLI you could run aptitude and it functions more or less the same as the package manager in a GUI.
2) You can use adduser or useradd to add a user to the system, usermod to modify any existing user accounts, etc. Groupadd, groupmod, etc do the same thing but for groups.
3) Server distros can be used for just about anything. The main differences between types of distros (server, desktop) are mainly in the packages included or installed by default.
4) Not installing updates by default is a good way to keep a potentially bugged update from screwing your system. Basically use whatever you want, but be cautious like with ANY automatic update system.
5) Yeah, most ISPs will block or at least filter traffic on the more commonly used ports for things like that.
6) Pretty much see my answer to #3.
I'm sure some in here may (or rather will) be able to explain any further, but that's pretty much "it".
Honestly, at least from personal experience over the past 5 years of using Slackware, Fedora, Arch and (failing to use) Ubuntu, it's important to search for a distro that has what you want. Slackware was my first distro and I still love it, though I wanted something a bit more modern, so I tried Ubuntu. It failed to work several times so I threw it aside and tried Fedora. Eventually I wanted something like Slackware but with the package management and "newness" of Fedora (which I had occasionally had some problems with). Thus I came to Arch and will probably stick with it for when I'm using Linux regularly again.
And when I talk about users I mean the casual users. My mother. My girlfriend. My friend, etc...
They don't care about licenses, they only want to use good software. If its pay software they will either pay or the more common route, they will pirate it. They will crawl shitty download sites, wait days for it to download, manage shitty keygens/cracks and all because its good software.
The majority of casual users probably aren't even aware that alternatives exist. You'd be amazed at the number of people that pirate Photoshop just for simple edits when Gimp would have sufficed or pirate Office when they could have done the same task in Open Office. There are people that are aware and prefer the proprieatary software for various reasons but a lot of them would be perfectly happy if they knew they could get the simaler versions that worked just as well for the same tasks.
We have to ask ourselves as linux users, what do we want? Do we stay closed in respect to licenses not allowing anything to go into distros in order to maintain the "purity" of linux like debian with linux-firmware? Or do we embrace the possibility of using proprietary software which collide with the open source movement in order to gain userbase?
I feel differently than most people on this point. For me I don't feel the FLOSS community is a numbers game. Having an extremely large userbase means nothing for FLOSS if the majority of those users just want proprietary software and it leads to stuff like Photoshop, Office, etc getting ports then that actually does damage to the community and movement imo. I'm sure some people would see that as a great thing though.
I always have been fond of the second one. The more users we got, the better our software will be, the more developers we attract, the more developers see that open source works and hopefully contribute to it.
Of course, then there is the other view of it. If there is propietary software that fills a void in linux, there is no reason to make an open source alternative because the proprietary one is what users use. No idea how to tackle that to be honest.
The problem you stated is one of the reasons why I'm opposed to it. Just think of the effort that would be going into a project like Gnash were the proprietary version of Flash not in wide use.
Oh, and I care about licenses. The only software I use on linux which is not gpl is skype (and it can die in a fire, fucking piece of shit). But I know the shortcomings of it, like not being able to play any game because of the experimental status of nouveau 3D. I am willing to put up with that and help the developers make a decent alternative to the nvidia driver. But as I said, casual users just want to play a game wich runs on linux, they don't care about the license of their display drivers. Its all about perspective and future
Fake edit: Im not being aggressive or anything, am I? Don't want to appear like that, as I said English is not my native language so excuse me if something sounds bad, not my intention
You're not not being aggressive at all. I actually enjoy these types of debates as I seem to have them very often. I fully understand how casual users feel about the current issues with drivers. I know most people don't care if it's a blob or not as long as they can just plug in their hardware and have it work well. I've done a few presentations in school on free software and it's always one thing that comes up. However as you stated previously this leads to less people using the non-proprietary versions which in turn leads to less testing being done which leads to lower quality drivers. The issue is partly due to the Linux kernel accepting the proprietary versions in the first place. Plenty of the kernel developers feel the same although it's unlikely to change anytime soon.
Okay so lets preface this a little bit. Sorry I'm such a newb in this area. My experience with severs is basically well 0. I know CentOS seems to be the go to linux server, and is basically the free offspring of RHEL which is the other big nix server distro is you want to pay. Beyond that if your from the other camp Debian seems to be where it's at since like RHEL it's the slow , stable, not quick to change, but just works of all the well debian based distros.
At my job, we have two Ubuntu servers, a CentOS server an openSuSE server and a bunch of openSuSE. CentOS gives us the most trouble, for what it's worth.
2) How do you manage users?
The install had me create my user name and password to login with, but I was sort of confused. Does that initial account I created mean it's the master say admin account? What about creating other accounts? How would you manage what users have access to and such (would you call this group policy?)?
You could try out webmin. It gives a centralised, web-based interface for many common server software packages as well as other aspects of the system, like user management, perl module management.... like everything.
5) Is it true you can't host your own server (say wordpress blog or ftp etc...) because the ISP will block you even if you can get it setup?
Like say you wanted to run a small time wordpress blog. Is it true you can't get the connection out even if you can get wordpress installed on a lamp stack? They just port block or something?
You can use a nonstandard port. I run a web proxy on port 888 in my house on a server roughly the size of my fist, but I could easily use that port for a web server.
6) Other than the GUI and what comes preinstalled depending upon the distro is there really any difference between a desktop and server distro?
I mean is the only difference between the two the fact that one boots directly to a terminal without the GUI overhead? Obviously there are differences in what's preinstalled, but you could install/uninstall to get the same packaged and stuff on both for them to be the same right? I mean the kernel itself is the same for both flavors of linux right?
That's most of the difference. Server kernels might be lighter than other kernels -- that is, either fewer things might be compiled into it, or there may be fewer kernel modules. But the difference is pretty darned minimal unless you're using some sort of edge case (like a server specifically built for an embedded system). You can choose to use a desktop kernel on a server, and you won't notice it.
On CentOS, we had to use a very specific kernel to get our machine's RAID working. Pain in the butt.
...hm. 16 minutes to my final. I should start studying.
That's what throws me from the command line terminal interface. How do you even know wtf is even in the repos beyond just guessing.
Also how do you go about adding in packages not found in the repos? On the desktop I immediately head to the web, but I'm sort of clueless on wtf to do on a server.
Also how do you go about adding in packages not found in the repos? On the desktop I immediately head to the web, but I'm sort of clueless on wtf to do on a server.
some options:
1. install X and a graphical web browser onto the server.
2. install elinks, a text-based web browser
3. do your package hunting on a workstation and scp the install files to the server.
That's what throws me from the command line terminal interface. How do you even know wtf is even in the repos beyond just guessing.
Also how do you go about adding in packages not found in the repos? On the desktop I immediately head to the web, but I'm sort of clueless on wtf to do on a server.
Speaking of a headless server... it's one of the very first things that I got downright intimate with before working with desktop linux. The easiest way to manage apps on an Ubuntu box is aptitude. Aptitude is like apt-get but on steroids. It basically links and creates any broken dependencies for any packages you go out and look for without having to mess with adding things to your source list. It also has an easy on the eyes UI that you can look at within a terminal on a headless server. BTW, if you haven't messed with adding sources to your source list, google that right now and read about it.
Ironically, am more comfortable doing things in terminal (and prefer it) to using a UI. While looking for things in your servers (be it keywords with grep or whatever the case might be) get used to looking at a terminal window instead of a UI or a box with drop downs. Me personally, I came from a Windows only world and just made the switch in February of this year. While it is a hard transition to make, it is completely worth it.
I would also highly advise against installing an X session on a server because that opens up a LOT of vulnerabilities and I think it does learning the terminal and lay of the land dis- justice. If you absolutely need a UI, install XFCE on an Ubuntu server. Very low specs and memory usage and is quite snappy.
The way I work is I've got either a Windows or Linux instance on my laptop running (meaning native in either/or) and I use SSH into my servers. At any given time I've got about 5-15 terminal windows up on my screens. Do yourself a favor and start to customize the terminal window screens by font color so you can quickly tell them apart. I also have background images on them for some added sauce. Right now I've got green/black as my workstation type, orange/black as my home SSH connection, yellow/black for production servers and blue/black for development boxes.
It's kinda funny as I type all of this out because it's only been a few months for me and I feel like I've got a good grasp of things. I'll answer your questions you posted above later tomorrow, right now I'm bushed. Happy hackin'!
FAKE EDIT-EDIT: To answer your question, Google, linuxforums, Ubuntuforums and stackoverflow should be on your bookmark lists from here on out if they aren't already. I knew webmin was canned a while back as I was looking for a good interface to look at/connect to from a web browser but it turned out... I don't need one. ;P
1) How can you tell what programs you can install?
So I ran my install, partitioned my drive, and everything went fine. Then it allowed you to pick from some big services people normally run or just go ahead. I decided fuck lets install the LAMP stack, and that worked fine. Set up my user name password etc... etc... and got to the terminal screen.
I was like well wtf now? This was your idea after all (talking to myself in my head LOL)! I decided lets try and install a few things. So I ran the normal sudo apt-get install myphpadmin and that ran fine. Did the same thing for wordpress. Ditto for Emacs.
I assume these all worked because they were already in the repository. Yet at the command line interface how the fuck are you suppose to know? In desktop say Ubuntu 11.04 or the Fedora 15 beta with Gnome 3 you have your package manager. You searched and it came up. If not you hit the interwebs for either a .deb/rpm, tar, or say a PPA for Ubuntu. There seemed to be know way to visually see the package manager so I feel like I just got lucky that the shit I tried to install was there.
apt-cache search whatever shows you a list of the packages/descriptions refering to the search. apt-cache search vns would show all the packages that have vnc in the name or in the description.
Also aptitude. Learn it. Love it. An evolution of the old dselect front-end for dpkg.
2) How do you manage users?
The install had me create my user name and password to login with, but I was sort of confused. Does that initial account I created mean it's the master say admin account? What about creating other accounts? How would you manage what users have access to and such (would you call this group policy?)?
3) Would the server distros really only be good for enterprise solutions?
Are the server distros just good for hosting large databases, and file storage and the like for large corporations? You know like running the London Stock Exchange? What about personal usage such as a media server, a backup server, or hosting your own site?
Basically for home usage is much better an easy to manage distro with up-to-date apps. Server distros are much more capped, have more issues to overcome and lack support for newer hardware to be that useful. You can always disable the X server on a normal distro so for personal use its easier to grab a normal one.
4) It appeared the top answer aka default for what to do with updates was manually manage on your own? Is this really the case?
The 2nd option was for automatically installing security updates and the like. Why wouldn't I want it to automatically update or at least auto notify me (I think this was the 3rd option) about updates? Just afraid updates might break shit? Wouldn't you want the security updates installed ASAP?
For once, bugs in updates. For seconds, sometimes there is really special configurations done or patches applied to programs which won't work on new versions. You don't want to have a call at 4:am a saturday because the web server stopped working after an automatic update
Also thanks to syslog-ng/rsyslog, you will get an email daily with all the changes to the syste, logs for services and packages which have updates so if a huge hole appears and a patch is released you will know that you can upgrade it.
5) Is it true you can't host your own server (say wordpress blog or ftp etc...) because the ISP will block you even if you can get it setup?
Like say you wanted to run a small time wordpress blog. Is it true you can't get the connection out even if you can get wordpress installed on a lamp stack? They just port block or something?
Seeing the other responses, it seems that the ISPs block ports in the US. Strange because In Spain I got nothing blocked. I run several services from home so I can still do things while at work.
6) Other than the GUI and what comes preinstalled depending upon the distro is there really any difference between a desktop and server distro?
I mean is the only difference between the two the fact that one boots directly to a terminal without the GUI overhead? Obviously there are differences in what's preinstalled, but you could install/uninstall to get the same packaged and stuff on both for them to be the same right? I mean the kernel itself is the same for both flavors of linux right?
Depends on the distro. For example Red Hat/Centos normally NEVER change to major versions during the product lifetime. That means that you could be running apache 2.0.X for 4 years until the next iteration comes and apache 2.2.x is into it.
Same with debian. They rather have older packages but stable than newer and exposed to new vulnerabilities.
Also hardware support is hit or miss. Red Hat 5/Centos 5 didn't have any kind of wireless support. They launched Red hat 6 4 months ago with wireless out of the box. It wasn't needed and the modules weren't stable enough to be included. You could install them but you were on your own.
Also there is more security features from the start like the firewall up and running blocking everything unless you specify it or apparmour enforced. Selinux policies as well, which were a pain in the ass because they were fucking awful to write and transform into modules.
hat's what throws me from the command line terminal interface. How do you even know wtf is even in the repos beyond just guessing.
Also how do you go about adding in packages not found in the repos? On the desktop I immediately head to the web, but I'm sort of clueless on wtf to do on a server.
In the end, after working with the terminal for some time you already know the name of the packages and such. So it's not really a problem. And normally you don't really add extra repositories out of a couple of well know ones (rpmforge for example) so its a non issue.
Ironically, am more comfortable doing things in terminal (and prefer it) to using a UI. While looking for things in your servers (be it keywords with grep or whatever the case might be) get used to looking at a terminal window instead of a UI or a box with drop downs.
Hell yeah, terminal is awesome. The only reason I have the xserver is because I can have several terminals open at the same time and the browser.
I mean, sometimes I even write the goddam emails from the terminal because is faster than opening the web browser. Not to mention opening every text file in nano for no reason.
mailto -s "I'll arrive shortly" somebody@something.com
I'll be there in 5, Im on my way.
EOF
The way I work is I've got either a Windows or Linux instance on my laptop running (meaning native in either/or) and I use SSH into my servers. At any given time I've got about 5-15 terminal windows up on my screens. Do yourself a favor and start to customize the terminal window screens by font color so you can quickly tell them apart. I also have background images on them for some added sauce. Right now I've got green/black as my workstation type, orange/black as my home SSH connection, yellow/black for production servers and blue/black for development boxes.
Ooohh, I like this idea. In the coming weeks I plan on setting up my own headless server, mostly for use as a permanent IRC connection and other small things. Seems like a fun project. I plan on SSH'ing in from my Android phone and laptop.
I think I'm going with Ubuntu Server to start with, because I'm a server noob. I assume I won't need to take any extra security measures after installing it?
Btw, should I use an IRC bouncer? Or just irssi? I'm not sure what the advantages of setting up your own bouncer are...
//Edit: so it seems that one difference between a screen+irssi session and a bouncer is that you can log in to the latter via your own IRC client, instead of through a terminal window. I guess that's an advantage, are there any others?
Okay so lets preface this a little bit. Sorry I'm such a newb in this area. My experience with severs is basically well 0. I know CentOS seems to be the go to linux server, and is basically the free offspring of RHEL which is the other big nix server distro is you want to pay. Beyond that if your from the other camp Debian seems to be where it's at since like RHEL it's the slow , stable, not quick to change, but just works of all the well debian based distros.
Anyways as I stated a week or two ago I got a free sata HDD from an old comp, and since I dual boot I decided to make that linux only. Since I had a ton of space and only had Ubuntu 11.04 desktop on there I got bored late last Saturday night, and decided I'd try out Ubuntu Server. Sure it's not one of the aforementioned distros. Yet it's a new version that just hit, and I'm sort of an Ubuntu guy.
I say this because some of this is general questions, and some of this might just be how Ubuntu handles shit. WTF do I know honestly.... So here it goes....
1) How can you tell what programs you can install?
So I ran my install, partitioned my drive, and everything went fine. Then it allowed you to pick from some big services people normally run or just go ahead. I decided fuck lets install the LAMP stack, and that worked fine. Set up my user name password etc... etc... and got to the terminal screen.
I was like well wtf now? This was your idea after all (talking to myself in my head LOL)! I decided lets try and install a few things. So I ran the normal sudo apt-get install myphpadmin and that ran fine. Did the same thing for wordpress. Ditto for Emacs.
I assume these all worked because they were already in the repository. Yet at the command line interface how the fuck are you suppose to know? In desktop say Ubuntu 11.04 or the Fedora 15 beta with Gnome 3 you have your package manager. You searched and it came up. If not you hit the interwebs for either a .deb/rpm, tar, or say a PPA for Ubuntu. There seemed to be know way to visually see the package manager so I feel like I just got lucky that the shit I tried to install was there.
As most of us have stated... aptitude on a headless server is how you manage the apps. The way to find out what you need is by Google/Bing searches on what you need. There really isn't much black magic to it as that is the world of open source.
2) How do you manage users?
The install had me create my user name and password to login with, but I was sort of confused. Does that initial account I created mean it's the master say admin account? What about creating other accounts? How would you manage what users have access to and such (would you call this group policy?)?
I'm still on the learning curve for this one on the headless server but the basic commands of adduser addgroup are the way to do it. The reason why it's important to learn these is because this is how you harden your server. Other times, when using a LAMP stack, you will give permissions to www-data which is the apache2 user to access the /var/www folder and have access to the files there outside of the provisioned "root" / "users" on the server. Making new users, giving them passwords and assigning them to groups is something I have become more familiar with but still need some more time to develop. In terms of calling this AD or GPO related, it's really not. Permissions are given based on where you want user to get to. You can also lock down applications using visudo and only assign certain users to use certain apps. Again, Google/Bing are your friends.
3) Would the server distros really only be good for enterprise solutions?
Are the server distros just good for hosting large databases, and file storage and the like for large corporations? You know like running the London Stock Exchange? What about personal usage such as a media server, a backup server, or hosting your own site?
Server distros can be anything you want them to be. I've got an old P4 1.7GHz w/ 256MB of RAMBUS ram running my Ubuntu home server and right now, it's currently using 35MB/243MB. Insane right? I mainly only use it for SSH and OpenVPN connections and don't use it for any file storage. Figuring out what you need is part of the fun of Linux in my opinion. Lots of people use Linux for HTPC's ala http://xbmc.org/ and rightly so. It's one of the better solutions for Media. Other people build a SAMBA server and run a home NAS. Why not? Other host custom home pages for their internal network at home or run a web site for someone on their server.... the sky really is the limit (as well as the ISP rules).
4) It appeared the top answer aka default for what to do with updates was manually manage on your own? Is this really the case?
Security updates get applied without user intervention. I mentioned looking at your source list. Go look at: /etc/apt/sources.list (so... sudo nano /etc/apt/sources.list) You'll notice that the release (main restricted) of your Ubuntu version listed there are commented by the ##Major bug, fix, updates produced after final release. That channel is what checks for updates from Canonical regardless if you do anything. Always have these updates apply. When I run updates on my machines I do two commands:
sudo aptitude update
sudo aptitude upgrade -v
The -v is for verbose. I want to see EVERYTHING that is being updated. If you're wanting to play it even MORE safe, you can run this command:
sudo aptitude safe-upgrade
What that one does (and this is why aptitude wins vs. apt-get) it will check other dependencies of the application against other installed apps and then use logic to decide whether or not that upgrade will break anything else on your system. Brilliant!
The 2nd option was for automatically installing security updates and the like. Why wouldn't I want it to automatically update or at least auto notify me (I think this was the 3rd option) about updates? Just afraid updates might break shit? Wouldn't you want the security updates installed ASAP?
Depends. Some ISP block certain ports as they have a clause that says you cannot run a home "server" on the standard ports. Test it and try it. Only way to find out.
Like say you wanted to run a small time wordpress blog. Is it true you can't get the connection out even if you can get wordpress installed on a lamp stack? They just port block or something?
You can always change the port that the server is hosted on in the apache2 config. >
6) Other than the GUI and what comes preinstalled depending upon the distro is there really any difference between a desktop and server distro?
I mean is the only difference between the two the fact that one boots directly to a terminal without the GUI overhead? Obviously there are differences in what's preinstalled, but you could install/uninstall to get the same packaged and stuff on both for them to be the same right? I mean the kernel itself is the same for both flavors of linux right?
IDK I feel like I have more questions, but I'll start with those for now. Time to go for my nightly walk to get some exercise! Thanks in advance!
Biggest difference is headless vs. head. Each distro out there has their own default install along (usually) with the options to install most common server choices such as SSH, LAMP, Tomcat, Mail server, DNS, SAMBA, etc. Also on each version or distro, kernel versions will differ for what versions of software they want to support.
Without getting into this too deep, imagine you have a software application that runs on PHP 5.2.x Any and all newer versions of the distro you are currently using have PHP 5.3.x. What is a sysadmin to do? This is why the LTS from Ubuntu is so key and moves along more at a snail's pace than a turtles pace (with the release versions of Ubuntu). Application compatibility is what drives linux distros and really nothing else. It took me a while to figure that out but it's the truth at the end of the day. Servers especially are dependent on certain versions of LAMP stacks to remain the same as companies have coded their sites specifically for that version of PHP or apache2 or Perl. Across the various distros, CentOS is the most akin to RHL and that's what most people like in the server space because it is solid, tired and true tested to just always work. Then there is me who hates Red Hat (RPM's for that matter) with a passion ever since my early days of working with RHL at Intel. Everyone is going to have their biases on what to use and how to use it, that's what makes the Linux world so interesting and engaging. Function over faction is a term that is tossed around quite a bit in the Linux world because at the end of the day, we're all on the same team regardless of which distro you prefer.
apt-cache search whatever shows you a list of the packages/descriptions refering to the search. apt-cache search vns would show all the packages that have vnc in the name or in the description.
Also aptitude. Learn it. Love it. An evolution of the old dselect front-end for dpkg.
You can use aptitude textually as well as ... ncursesly. "aptitude search <packagename>" gives similar, but not exactly the same, results as "apt-cache search <packagename>", but they're in a prettier format. It's also nice to be able to use the same command for searches as for installs "aptitude install <packagename>".
However, I recommend installing "apt-file", which I have found useful occasionally. If you've ever said "hey, this thingy I need to use says it requires "/usr/lib/libwine.so.1", but I have no idea what package has that!", then you can just type "apt-file search /usr/lib/libwine.so.1".
Seeing the other responses, it seems that the ISPs block ports in the US. Strange because In Spain I got nothing blocked. I run several services from home so I can still do things while at work.
Much of the reason for port blocking comes in part, imo, from the misbehavior of SMTP as a malware vector. For some reason, some company made a mail program that was so impossibly insecure that summoning a message in a message preview pane could cause code embedded in the message to auto-send copies of itself to everyone in that user's address book. This strangled bandwidth a bit and allowed malicious software to place itself into millions of computers in an astoundingly quick time.
Something similar happened using incredibly wide open exploits from a semi-popular web server which could be installed by regular joes. Anyway, one reason why ports were closed was to dramatically slow this sort of quick infection. Not just from these specific programs, but from any server app which was listening on a computer that would otherwise be accessible from anywhere on the Internet. Of course, everyone using NAT (routers) solved many of the same problems anyway.
To push this even more: It's a tabbed web browser with menus and colours and support for various protocols, optional javascript support (apparently), a download manager, mouse support, bookmarks, session saving.... It's pretty much as good as you can get in a web browser without involving graphics.
Hell yeah, terminal is awesome. The only reason I have the xserver is because I can have several terminals open at the same time and the browser.
screen -Rd <somenameyoumakeup> is the best tool in the universe for managing terminal instances. It's a tabbed terminal manager that runs entirely within a terminal. It is highly configurable and includes the magic technology of detachable sessions. If you have a bunch of text apps doing stuff in a bunch of screen tabs, and you decide to play some DOOM or something, and something goes wrong and your whole login session detonates, all that stuff in screen will still be running, and you can just reattach it to any vtty or X terminal app (or over ssh -- you can start work at the server then detach and reattach from your workstation!).
I mean, sometimes I even write the goddam emails from the terminal because is faster than opening the web browser. Not to mention opening every text file in nano for no reason.
Thanks for all the awesome replies. You guys have given me alot of great knowledge. I have another 6 questions, but I'll probably wait and pose them tomorrow (or later when I'm home from work).
You can use aptitude textually as well as ... ncursesly. "aptitude search <packagename>" gives similar, but not exactly the same, results as "apt-cache search <packagename>", but they're in a prettier format. It's also nice to be able to use the same command for searches as for installs "aptitude install <packagename>".
However, I recommend installing "apt-file", which I have found useful occasionally. If you've ever said "hey, this thingy I need to use says it requires "/usr/lib/libwine.so.1", but I have no idea what package has that!", then you can just type "apt-file search /usr/lib/libwine.so.1".
Never used apt-file, will install it because that sounds handy. It actually happened that I needed some libraries but I usually search for libxxx-dev as usual to find them.
Also, something that I forget to mention is that red-hat based systems had a nifty feature called groupinstall in order to install a group.
So if you needed to install a web server you would do a "yum groupinstall "Web server" " and it will install everything needed. Another way of installing things from the terminal.
I think in debian based distros the same is achieved trough meta-packages like xfc4 which depends on all packages that compose the xfce4 desktop.
screen -Rd <somenameyoumakeup> is the best tool in the universe for managing terminal instances. It's a tabbed terminal manager that runs entirely within a terminal. It is highly configurable and includes the magic technology of detachable sessions. If you have a bunch of text apps doing stuff in a bunch of screen tabs, and you decide to play some DOOM or something, and something goes wrong and your whole login session detonates, all that stuff in screen will still be running, and you can just reattach it to any vtty or X terminal app (or over ssh -- you can start work at the server then detach and reattach from your workstation!).
Thing is, I normally work on the 4 terminals at once so I have to see them all at the same time. Terminator is awesome, love how you can split the terminal vertically or horizontally.
You misspelled "joe". Easy mistake to make, I see it happen all the time.
I refuse to use any text editor which involves using the ^ key. Its a pain in the ass in Spanish keyboards. Control + whatever is much easier and I can do with my left hand almost everything except cut (control+U)
At least you didn't say vi or emacs
Thanks for all the awesome replies. You guys have given me alot of great knowledge. I have another 6 questions, but I'll probably wait and pose them tomorrow (or later when I'm home from work).
Post them rigth away! Im on my way to work so Im gonna have the rest of the night free unless a server in Australia decides that its time to die. Stupid Equinix and their shitty lines always give problems :S
So if you needed to install a web server you would do a "yum groupinstall "Web server" " and it will install everything needed. Another way of installing things from the terminal.
zypper pattern (to list patterns, which is their name/equivalent for groups) zypper info -t pattern <pattern_name> (to list info on the pattern, including the packages it installs) zypper install -t pattern <pattern_name> (to install all packages in a given pattern)
In Arch, it's like this:
pacman -Sg (Sync / groups -- gives a list of all groups) pacman -Sg <group_name> (Sync / group -- shows all packages belonging to <group_name>) pacman -S <group_name> (Sync -- installs (synchronizes with repo) packages in <group_name>)
Mandriva's package manager is ridiculously powerful, but it's been too long since I last played with it. Gentoo is supposed to be amazing in this regard, allowing you to apply "USE" flags, iirc, in a way that it will always and/or never install stuff from certain groups based on your specific criteria -- like, if you never wanted java apps installed but you installed packages that list them as optional dependency, you can use set it to behave in that fashion. This is a different thing from our patterns/groups, but I thought it related enough to point out.
For what it's worth, though, I've only installed a pattern/group on the command line once in history ("pacman -S lxde", about a month ago).
Thing is, I normally work on the 4 terminals at once so I have to see them all at the same time. Terminator is awesome, love how you can split the terminal vertically or horizontally.
screen does that, too. Well, it does vertical splitting, at least. Nonetheless, it is still a nice advantage to use screen (without using the tabbing capability, so you get just one shell session) inside each of your terminal instances in order to take advantage of the session persistence.
I refuse to use any text editor which involves using the ^ key. Its a pain in the ass in Spanish keyboards. Control + whatever is much easier and I can do with my left hand almost everything except cut (control+U)
Joe uses the control key. When you see something like "^C" listed in documentation, it just means "Hold the control key and hit the C key". There's no actual usage of the carat symbol itself in joe. Or am I misunderstanding what you're saying?
Okay I have a few more questions.... thanks for the replies so far, and not taking me as being to pushy or n00bish! You guys rock!
7) What processes start up when you 1st boot? Do you have to manually start everything back after a reboot? Do you just have some script written to start all of your processes back up?
8) Do you have to be logged in for stuff to be running?
Like if I'm using the server for say back up can I be logged out, and the backups still happening? If I'm running say a website and using the server to host do you have to be logged in for the site to be up, or does the server just need to be booted up?
9) Do you guys and gals do all of your work on the server itself or do you all just remotely log in from say a workstation versus being on the actual box?
10) If someone else logs in with different privileges does that mean whatever processes you had started aren't running anymore?
11) Is SSH where it's at for doing remote logins? Do you have to setup SSH for each user?
12) Has anyone used Ubuntu Advantage more specifically Landscape(or the equivalent for other distros?)?
Shit looks pimp as a mother fucker especially for dealing with large groups of workstations and servers. The problem is of course Canonical is gonna make it appear to be pimp. That's what they are trying to sell you on to make back some cash, and stay afloat. So who knows how it is in practice. I know they supposedly redid the front end like around late last Dec, but I have no experience with this obviously.
PS: I got more questions, but I figure I'll keep then in smaller groups and let people talk it out! Thanks in advance Linux crew!!
Gentoo is supposed to be amazing in this regard, allowing you to apply "USE" flags, iirc, in a way that it will always and/or never install stuff from certain groups based on your specific criteria -- like, if you never wanted java apps installed but you installed packages that list them as optional dependency, you can use set it to behave in that fashion. This is a different thing from our patterns/groups, but I thought it related enough to point out.
Tried gentoo years ago and desisted because everything took hours to compile on my shitty machine.
Should give it a go now that I have an awesome desktop!
screen does that, too. Well, it does vertical splitting, at least. Nonetheless, it is still a nice advantage to use screen (without using the tabbing capability, so you get just one shell session) inside each of your terminal instances in order to take advantage of the session persistence.
Oh, didn't know that. Time to ream the man pages for it to see if it can replace my love for terminator.
Joe uses the control key. When you see something like "^C" listed in documentation, it just means "Hold the control key and hit the C key". There's no actual usage of the carat symbol itself in joe. Or am I misunderstanding what you're saying?
7) What processes start up when you 1st boot? Do you have to manually start everything back after a reboot? Do you just have some script written to start all of your processes back up?
on the rest there is soft links to the main programs.
The scripts inside are in this type:
S10name
Where the first letter is S for start or K for kill. The second number is the priority, the lower the number the sooner the service start (We don't want to start the web server before the network, do we?) and the rest is the name of the script it has to start or stop.
The directory names make reference to the different runlevels of a linux machine, which are 7
0 - Halt mode
1 - Single user mode
2 - Multi user mode (no networking)
3 - Full multi user mode
4 - not used
5 - Xserver multi user mode
6 - Reboot mode
As you can see, we don't need the same services for a single user mode than a Xserver user mode. And in the case of Halt mode, all the scripts start with a K because we need to stop everything to poweroff the system.
You can enable and disable services by managing the different links in the rcX folders. For example, I think that in Ubuntu runlevel 4 and 5 are the same one, xserver + gdm multiuser mode. You could change the scripts on rc4.d in order to have it NOT launch GDM and launch KDM instead.
Or make a runlevel with everything but networking and xfce4 running. Or whatever, I don't know!
In order to manage the scripts there are some utils. In ubuntu you have update-rc.d to enable or disable services on different runlevels.
So if you installed apache2 but want to start it manually instead of having it run always, because you just want to play with webpages from time to time, not all day, you would remove it doing:
sudo update-rc.d -f apache2 remove
In order to reenable it again with the default runlevels (we don't want it to start when rebooting. You can actually start it on runlevel 6 if you want LOL) you will run:
sudo update-rd.c apache2 defaults
P.S.: Different distros manage this differently. I tackled ubuntu because you are using it, but my explanation up here could be a bit imprecise. I use centos mainly so excuse me if I mistaked a command here or there
8) Do you have to be logged in for stuff to be running?
Like if I'm using the server for say back up can I be logged out, and the backups still happening? If I'm running say a website and using the server to host do you have to be logged in for the site to be up, or does the server just need to be booted up?
See above. The services start by themselves depending on the runlevel. No need for anyone to be logged.
9) Do you guys and gals do all of your work on the server itself or do you all just remotely log in from say a workstation versus being on the actual box?
I do everything on remote. We used to do it with ILO cards (Integrated ligths out) which was a computer inside the computer (YO DAWG), and allowed us to stop or start the servers like we were there and we never losed the connection or the screen.
10) If someone else logs in with different privileges does that mean whatever processes you had started aren't running anymore?
Nope. Services are launched with their own privileges and users. Apache is normally launched witht he user/group www-data. This users are normally created automatically and have very restricted privileged and they can't even login in a shell.
11) Is SSH where it's at for doing remote logins? Do you have to setup SSH for each user?
ssh is the more secure one and easy to implement. You set up ssh and every user can log in as long as they have an account on the machine,
12) Has anyone used Ubuntu Advantage more specifically Landscape (or the equivalent for other distros?)?
Shit looks pimp as a mother fucker especially for dealing with large groups of workstations and servers. The problem is of course Canonical is gonna make it appear to be pimp. That's what they are trying to sell you on to make back some cash, and stay afloat. So who knows how it is in practice. I know they supposedly redid the front end like around late last Dec, but I have no experience with this obviously.
Nope but I would love to. I guess is something like vmware esxi where you can see everything for every server and manage them remotely, which is pimp as fuck.
Okay I have a few more questions.... thanks for the replies so far, and not taking me as being to pushy or n00bish! You guys rock!
7) What processes start up when you 1st boot? Do you have to manually start everything back after a reboot? Do you just have some script written to start all of your processes back up?
scroll down to the bolded TL;DR if you want a massively compressed version of this following wall of text.
Okay, the kernel (the part of the OS that is actually "Linux", sort of a single program with plugins that does the most low-level stuff, like drivers and memory management and so forth) starts a program called "init". init looks at a text file in /etc/ (/etc/rc in BSDs, /etc/inittab in most Linuxes). /etc/rc is just a shell script, but /etc/inittab is more like a config file that tells init how to proceed. I'll just give you information on the latter.
From here, what happens entirely depends on what is in inittab. You can start up your machine (controlled with the key combo Ctrl-Alt-F1 through Ctrl-Alt-F8) with eight separate text terminals or eight full graphical sessions or any combination thereof, each on its own virtual screen. Most Linux-based systems start with six full-screen, text-only terminals (Ctrl-Alt-F1 through Ctrl-Alt-F6) and one X11 (graphical) session (Ctrl-Alt-F7). X11 has the ability these days to spawn onto additional virtual screens so that multiple people can be logged in at the same time. You could actually do this sort of user switching in X11 about twenty-five years ago, though it wasn't implemented in as quite a user-friendly fashion.
I don't know why they have so many terminals. One or two would be fine for troubleshooting, but six seems like needless overkill. Last time I used Ubuntu, I discovered that this particular version had zero text-only terminals (or maybe it somehow only spawns the terminals after it starts X11 or something). This is a terrible, shitty mistake, because Ubuntu proceeded to screw up on the nvidia drivers, and recovering the system became a pain in the ass.
I'm going to give you a needlessly complex rundown of the startup sequence of two distros -- openSUSE and Arch. itxaka
(if he or she has a baby it should be named mini-itxaka)
, gave a good end user style rundown so you know where things are, but I want to trace for you the path the actual OS takes.
_openSUSE_
init looks at /etc/inittab. It finds the line "id:5:initdefault:". This means that the default runlevel is 5. What does this mean?
Well, later in the inittab, you see lines like this: l0:0:wait:/etc/init.d/rc 0
l1:1:wait:/etc/init.d/rc 1
l2:2:wait:/etc/init.d/rc 2
l3:3:wait:/etc/init.d/rc 3
#l4:4:wait:/etc/init.d/rc 4
l5:5:wait:/etc/init.d/rc 5
l6:6:wait:/etc/init.d/rc 6
The second ("runlevel") column, which I've made bold for clarity, is the runlevel. The third ("action") column ("wait", in all these cases) just tells init that it's going to run a program and wait for it to end before proceeding (instead of, say, forking it into the background and moving on). The last column is a program/script that it runs (/etc/init.d/rc <somenumber>). Basically, all that's going on with any of these lines are that any time it reaches the line's runlevel, it will run the program at the end of the line, waiting until that program is done.
There's also this line: ca::ctrlaltdel:/sbin/shutdown -r -t 4 now
This only works in terminal mode, not in X11, but it's a special action that works in any runlevel. If Ctrl-Alt-Del is pressed in any runlevel, it runs shutdown -r to reboot safely.
Around the end, you see these lines: 1:2345:respawn:/sbin/mingetty --noclear tty1
2:2345:respawn:/sbin/mingetty tty2
3:2345:respawn:/sbin/mingetty tty3
4:2345:respawn:/sbin/mingetty tty4
5:2345:respawn:/sbin/mingetty tty5
6:2345:respawn:/sbin/mingetty tty6
They mean that in runlevels 2,3,4 or 5, init will run /sbin/mingetty tty<somenumber> with the "respawn" action, which means that init will restart that process if it exits or crashes. For what it's worth, the mingetty commands are those text-only terminals in Ctrl-Alt-F1 through Ctrl-Alt-F6.
tl;dr: When the kernel has done all its magic voodoo low level stuff, it calls init. Init looks at the inittab file. "id:5:initdefault:" tells it to start at runlevel 5, so it just runs any program specified in inittab where the second column includes a 5, which will be: l5:5:wait:/etc/init.d/rc 5 as well as all those mingetty things above.
/etc/init.d/rc is a script, and all it basically does is -- in openSUSE's case:
(A) runs "/etc/init.d/before.local" if it exists (this is a file that you can add commands to if you want stuff to run before anything else)
(B) If the runlevel is changing (eg, this is not bootup), it finds all scripts in /etc/init.d/<oldrunlevel> starting with the letter 'K'. Each of those safely kills a service/daemon.
(C) Finds all scripts in /etc/init.d/<runlevel> starting with the letter 'S' and runs them. Each of those start a service/daemon.
(C) runs "/etc/init.d/after.local" if it exists (again, a script you can put your own commands into)
It does other stuff to handle special cases, but that's basically the gist of it. Of special note: "/etc/init.d/rc5.d/" contains all the services that are started when the system boots (because it starts in runlevel 5, right?). One of the last things in there is "S10xdm". That's the script that starts X11.
_Archlinux_
init looks at /etc/inittab. It finds "id:5:initdefault:", so it knows that we're booting at runlevel 5. The line "rm:2345:wait:/etc/rc.multi" tells it that when we're in runlevel 5, it runs the "/etc/rc.multi" script (I'll get to that in a moment). It also has six lines for respawing "agetty" commands -- they do they same thing as the "mingetty" commands in openSUSE.
Then, at the very end, there's a line like this: x:5:respawn:/usr/sbin/lxdm
That line starts X11 with LXDE's display manager. LXDE isn't GNOME, so it's good to use in your graphical sessions.
;P
So when init looks at inittab on Arch's bootup, it learns that "5" is the bootup runlevel and it runs /etc/rc.multi, starts six terminals and then starts lxdm. Note that it starts up the graphical stuff right in the display manager instead of using a separate script to start it. If you don't want X11, you could just comment out that line, and if you wanted to use a different login manager (like slim or kdm), you could just edit that line.
Anyway, Arch has a configuration file called /etc/rc.conf. Among many, many other things, this is where you specify the services/daemons you want to start at runtime. Here's the relevant line on my computer: DAEMONS=(syslog-ng network netfs autofs crond dbus sshd ntpd cups samba !ushare !mediatomb !webmin !authdaemond squid !courier-imap)
The scripts that start and stop the services/daemons are all in "/etc/rc.d". There's no 'S' or 'K' involved here, because Arch doesn't really see the point in changing runlevels when you're already booted up. rc.multi takes this list and runs them in order (it runs "/etc/rc.d/syslog-ng" then "/etc/rc.d/network", and so on). Anything not listed on this line won't run. Anything with a '!' before it will be explicitly disabled on bootup.
_Notes_
The benefit of the more complicated method (I believe that Ubuntu/Mandriva/Fedora/etc closely resemble what openSUSE does) is that it's more flexible. For example, the /etc/init.d/rc script in openSUSE has some special voodoo that can allow services to start in parallel. nfs file sharing can sometimes take a few seconds to start, so it'd be great if you could start up the sound card handler and the bluetooth stuff at the same time. You can do that on openSUSE but not on Arch.
The benefit of Arch's way is that it's insanely easy, doesn't make your head hurt and works totally amazingly well for 99% of situations.
----- TL;DR: /etc/inittab is where different versions of Linux start being different. Most of them use "/etc/init.d" to contain a list of their services/daemons. Most of then have directories labeled "rc0.d" through "rc6.d" (also "rcS.d", which is the single-user "oh shit I destroyed the computer and need to emergency recover everything!" startup). Those "rc" directories are inside /etc/init.d/ in openSUSE but just in /etc/ in Ubuntu. They contain symbolic links to the services/daemons in /etc/init.d which are named in such a way to tell the system in what order they start and stop when you're booting up or shutting down. The graphical stuff is usually the last service/daemon to start.
Also, Arch rules.
8) Do you have to be logged in for stuff to be running?
Like if I'm using the server for say back up can I be logged out, and the backups still happening? If I'm running say a website and using the server to host do you have to be logged in for the site to be up, or does the server just need to be booted up?
What itxaka said, but I'd like to add that you often have the option of starting a server app while logged into your X11 session. In this particular case, the server app would exit when you logged off. But normally, servers start independently of user logins. Also, even as a user, you can tell programs to run completely independently of your login. That's what apps like "screen", "at" and "batch" are for, among others.
9) Do you guys and gals do all of your work on the server itself or do you all just remotely log in from say a workstation versus being on the actual box?
ssh is my master.
Access terminal on server? ssh myname@myserver
Access terminal from server where I have the same username as on my workstation? ssh myserver
Run aptitude to manage Debian or LinuxMint packages with arrow keys and stuff and then disconnect when done? ssh -t myserver aptitude
Restart a server app (this changes a little depending on your distro)? ssh -t root@myserver "/etc/init.d/servername restart"
or possibly ssh -t myserver "sudo /etc/init.d/servername restart"
Run some graphical program that's installed on the server? ssh -X myserver somegraphicalprogram
Run a multitabbed terminal that I can detach and resume from any workstation at any time? ssh -t myserver screen -Rd somenameyoumakeup
Got a video file on your server that you want to play on your workstation? ssh myserver "cat /path/to/porn.avi" | mplayer -
Think that mplayer is the spawn of the devil? ssh myserver "cat /path/to/porn.avi" | vlc -
Copy a file from your local machine to the server (using ssh)? scp /path/to/myfile myserver:/destination/path/
Copy web server config from one server to another (using ssh)? rsync -aPz --del myserver:/etc/apache2/ otherserver:/etc/apache2/
(note: 'P' means it gives you progress bars and stuff, 'z' enables compression during transfer)
Want to set up incredibly secure passwordless connections in ssh? ssh-keygen -t rsa
<follow instructions>
cat ~/.ssh/id_rsa.pub | ssh myserver "mkdir ~/.ssh; cat >> ~/.ssh/authorized_keys; chmod -R 700 ~/.ssh"
(the above's a little more than what's needed, but it makes sure that the server's ~/.ssh directory is present and secure)
Here's a script I wrote that will tell me the temperature in my server room: tempC="$(ssh topsecretservername "upsc S1500XLM" | grep temp | awk '{print $2}')";
tempF="$(/usr/bin/units -t "tempC($tempC)" "tempF")"
tempC="${tempC#0}"
tempK="$(math "$tempC + 273.15")"
echo -en "Server room temp:\n${tempK}K (${tempC}°C, ${tempF}°F)"
For incredibly nerdy reasons, it outputs the result in Kelvins, among other units. I just type "upstemp" in a terminal window, and it lets me know. I could also embed the command into a desktop applet so it continuously reports the server room temperature and, like, turns red or something when it gets too high.
Notes:
* adding -C after ssh will enable compression which might speed things up. The 'z' parameter in the above rsync example does the same thing
* there's also tunneling and other advanced stuff, but that's too scary for me. But I did play with reverse tunneling once (accessing a server that's behind a firewall by basically having it constantly try to contact your client machine -- say, at home -- and keep asking "Do you want me yet?" until the client machine decides to connect)
* Alt-F2 is the "Run Command" hotkey for the more popular desktop environments (KDE, LXDE, etc). I have a server at home that runs certain applications, so I made a very simple script that will just call ssh -X with a command name that I give it. I hit Alt-F2 and type "runfromhome pidgin" or "runfromhome medit" and the apps come up. I'm writing this right now on a graphical text editor (I don't trust writing long messages in forums on web browsers) that happens to be twenty miles away from my keyboard. I also have icons that I could click on to start them, but it's faster to type.
* You can also use ssh with ftp-like programs as well as mounting a directory on a server to a local directory as if it were a file share (using sshfs). Some file browsers, like KDE's Dolphin, will allow you to directly access files on your server by typing something like fish://myserver/ or sftp://myserver/ in the location bar.
10) If someone else logs in with different privileges does that mean whatever processes you had started aren't running anymore?
Multiple users can be logged in at once. Things started as services/daemons will of course, not be affected, and processes started by a user directly in their login session will only stop when they explicitly log out (if they didn't use "screen" or "batch" or "at"). So my logging in won't affect your stuff at all.
PS: I got more questions, but I figure I'll keep then in smaller groups and let people talk it out! Thanks in advance Linux crew!!
Since the OP is pretty old, what is the best distro to use for netbooks these days? Is it still Moblin? My netbook is pretty old (one of the first few that came out, a Dell Mini 9) and it only has 1 GB of RAM. I was also thinking of trying Ubuntu and Mint.
Since the OP is pretty old, what is the best distro to use for netbooks these days? Is it still Moblin? My netbook is pretty old (one of the first few that came out, a Dell Mini 9) and it only has 1 GB of RAM. I was also thinking of trying Ubuntu and Mint.
There are some suiting distros out there, I can totally recommend Lubuntu for an older netbook or Crunchbang, if you want to keep it puristic and stick to Debian instead of the *buntus.
Wow, I don't know then why I had that impression. I know that Red Hat started "obfuscating" their kernel patches because Suse was taking them and using them in their distro so I assumed they were almost identical.
I know that they both have that firsttime boot screens in order to configure everything after boot which are identical.
I also remember using yast from the terminal and being impressed of how easy it was to manage the system, but it remembered me way too much of red hat config utils on terminal as well, mainly the curses ui to the firewall.
^ has been shorthand for "control" since before windows, linux, dos or mac os and also probably since before either of us were born.
Isn't the @ in Arch used to background the starting services? It kinds of substitute that, you could use @bluez @alsa and they will both start almost at the same time and go to the background so they don't take boot time. Or are you talking about parallel starting of services? In that case, we only need to wait for systemd to take over all distros (except ubuntu I guess? they just changed to upstart so...)
For example I tunnel all of my outbound E-mail traffic back to my personal server to avoid having to change SMTP servers, use SMTP-AUTH, etc. when I am behind firewalls. I find that hotel firewalls, wireless access points, and the other various NATing devices you end up behind while traveling often do not play nice.
To do this I use the following:
ssh -f user@personal-server.com -L 2000ersonal-server.com:25 -N
The -f tells ssh to go into the background just before it executes the command. This is followed by the username and server you are logging into. The -L 2000ersonal-server.com:25 is in the form of -L local-port:host:remote-port. Finally the -N instructs OpenSSH to not execute a command on the remote system.
I haven't tried it yet, didn't have the need thought so maybe its more difficult than that.
Aaaaand... I just learned that you can create a local proxy that would tunnel everything trough your ssh'd server.
ssh -D 8080 -Nf example.com
Then you set up the proxy in your app as the local machine wiht the port specified and you get a secure connection being routed to your server hidden from anyone. I have to try this, I guess putty will do the trick.
There are some suiting distros out there, I can totally recommend Lubuntu for an older netbook or Crunchbang, if you want to keep it puristic and stick to Debian instead of the *buntus.
There are some suiting distros out there, I can totally recommend Lubuntu for an older netbook or Crunchbang, if you want to keep it puristic and stick to Debian instead of the *buntus.
I just went to http://www.android-x86.org/download and grabbed the Android iso there and started playing with it inside an emulator (VirtualBox -- by the way, VirtualBox and NIS don't work very well together -- I had to run it as root because it didn't believe me when I told it that I was part of its special "vboxusers" group).
Is this a good distro to use for netbooks? I have incredibly little experience with it, so I don't know.
Wow, I don't know then why I had that impression. I know that Red Hat started "obfuscating" their kernel patches because Suse was taking them and using them in their distro so I assumed they were almost identical.
I know that they both have that firsttime boot screens in order to configure everything after boot which are identical.
I also remember using yast from the terminal and being impressed of how easy it was to manage the system, but it remembered me way too much of red hat config utils on terminal as well, mainly the curses ui to the firewall.
...SUSE was derived from Slackware. Similarities to Red Hat probably come from the fact that both specifically target the Enterprise primarily, so they needed similar kinds of tools. That said, Yast is amazingly, astoundingly superior to any kind of similar attempt at a tool that Red Hat may have tried to make. The only centralized administration tool that I consider on the same level would be Mandrake's Control Center, but it's just not as good as it used to be (and while it has a curses interface, it's only for some of the modules).
Oh, webmin is pretty cool, too, but that's not distro-specific.
Also, wow, I had no idea that you could embed hyperlinks inside svg files. This makes html imagemaps even less useful than ever!
edit: And you can find-as-you-type on an svg!!! :O
And I thougth I was getting too technical with my explanation. Goddamit
I tend to do my best learning when it's answering a question that someone else poses.
Isn't the @ in Arch used to background the starting services? It kinds of substitute that, you could use @bluez @alsa and they will both start almost at the same time and go to the background so they don't take boot time. Or are you talking about parallel starting of services? In that case, we only need to wait for systemd to take over all distros (except ubuntu I guess? they just changed to upstart so...)
Mind blown. But, yeah, I was talking about parallel starting (based on the philosophy of a Makefile, apparently), but @ pretty much covers most of that advantage.
Next linux kernel release to be 2.8 instead of 2.6.X
From: Linus Torvalds <torvalds-AT-linux-foundation.org>
To: Linux Kernel Mailing List <linux-kernel-AT-vger.kernel.org>, linux-arch-AT-vger.kernel.org, DRI <dri-devel-AT-lists.freedesktop.org>, linux-fsdevel <linux-fsdevel-AT-vger.kernel.org>, linux-mm <linux-mm-AT-kvack.org>
Subject: (Short?) merge window reminder
Date: Mon, 23 May 2011 12:13:29 -0700
Message-ID: <BANLkTi=PLuZhx1=rCfOtg=aOTuC1UbuPYg@mail.gmail.com>
Cc: Greg KH <gregkh-AT-suse.de>, Andrew Morton <akpm-AT-linux-foundation.org>
Archive-link: Article, Thread
So I've been busily merging stuff, and just wanted to send out a quick
reminder that I warned people in the 39 announcement that this might
be a slightly shorter merge window than usual, so that I can avoid
having to make the -rc1 release from Japan using my slow laptop (doing
"allyesconfig" builds on that thing really isn't in the cards, and I
like to do those to verify things - even if we've already had a few
cases where arch include differences made it less than effective in
finding problems).
And judging by the merge window so far, that early close (probably
Sunday - I'll be on airplanes next Monday) looks rather likely. I
already seem to have a fairly sizable portion of linux-next in my
tree, and there haven't been any huge upsets.
So anybody who was planning a last-minute "please pull" - this is a
heads-up. Don't do it, you might miss the window entirely.
Did I miss any major development mailing lists with stuff pending?
Linus
PS. The voices in my head also tell me that the numbers are getting
too big. I may just call the thing 2.8.0. And I almost guarantee that
this PS is going to result in more discussion than the rest, but when
the voices tell me to do things, I listen.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
It amuses me highly that we've gone from a compatibility.major.minor versioning structure to some sort of random never.ijustfeellikeit.anychange versioning structure. Keeping track of software was much smoother when it did the former.
It amuses me highly that we've gone from a compatibility.major.minor versioning structure to some sort of random never.ijustfeellikeit.anychange versioning structure. Keeping track of software was much smoother when it did the former.
I guess is the number fight, like when...Madrake? bumped 2 versions to be on par with Red Hat. Or the Xbox 360 numbering in order to be on par with the Ps3.
I guess is the number fight, like when...Madrake? bumped 2 versions to be on par with Red Hat. Or the Xbox 360 numbering in order to be on par with the Ps3.
The thing is, it's really, really useful if all the versions where the first number is the same are mutually compatible to each other. All versions grub 1.x and all versions grub 2.x are pretty much entirely different programs, and the former can continue to be improved even as the latter matures. This worked incredibly well for apache, where apache 1.x prospered for many years after apache 2.x slowly developed into its own kind of awesome. KDE 3.x (under the project name Trinity) is still under development, even thought KDE 4.x has been mature for a very long time. Gnome 2.x will probably see continued attention while people adjust to their ... hesitation regarding Gnome 3.x.
It would be a clusterfuck if we used year.month in all those scenarios, because the newer, entirely rewritten versions are sometimes introduced in the middle of a year, and the older versions continue getting developed after their introductory year.
What would have happened if, in the middle of 2008, the new Grub came out?
We'd have Grub 2008.05 being completely incompatible with Grub 2008.06. Grub 2008.05 would continue to get upgrades and patches. What do we call the new versions? If they stay with the yyyy.mm scheme, then we get intersecting version numbers, where Grub 2009.02 is actually an improved Grub 2008.06 while Grub 2009.03 is actually an improved Grub 2008.05. Or maybe we'll call it Grub 2008.05.1, but that pretty much annihilates the point of yyyy.mm, since even though this particular Grub is mature and up to date (and intended for current usage), it is given an older version number that will confuse people. And some people might think that this newer version is just a day newer than what it's replacing.
Now, I guess you could do something like Major.Year.Month, but there are other aspects to non-dated release versioning. For example, the third (or sometimes the fourth) section, in some schemes, will explicitly tell you that we're dealing with security fixes and not feature upgrades.
A frequent updater of software might know that upgrading from killerapp 1.0.0 to killerapp 2.0.0 represents a complete rewrite of the software, upgrading from killerapp 1.0.0 to killerapp 1.1.0 represents an addition of features, and upgrading from killerapp 1.0.0 to killerapp 1.0.1 represents patching of holes and bugs. This gives you a lot more information than other schemes which only tell you "higher number means it's newer", and it makes it far less confusing when you have to decide whether or not to include a particular upgrade to your existing repository (debian stable, for instance, should generally only update the version increases that are security/bug fixes).
Edit: It doesn't seem altogether a bad idea to use a yyyy.mm versioning scheme for software suites/collections, though
Fedora installed and, well, it's different from what I'm used to from Arch and Ubuntu (I've actually only used those two distributions). I'll start with a few stupid, newbish questions since I'm heading to bed right now:
- I let Fedora install a new boot loader for me, but it only detected my Windows partition and not the Ubuntu one that I kept. Tomorrow I'll add the Ubuntu boot option in menu.lst, but if I want to return to the Ubuntu Grub 2-loader, can anyone give a few pointers on what to do? I've tried to understand Grub 2 but always failed...
- I don't really feel like reinstalling Matlab and some other programs, can I just symlink those from my Ubuntu partition once it has a fixed mount point, and run them like usual?
Fedora installed and, well, it's different from what I'm used to from Arch and Ubuntu (I've actually only used those two distributions). I'll start with a few stupid, newbish questions since I'm heading to bed right now:
- I let Fedora install a new boot loader for me, but it only detected my Windows partition and not the Ubuntu one that I kept. Tomorrow I'll add the Ubuntu boot option in menu.lst, but if I want to return to the Ubuntu Grub 2-loader, can anyone give a few pointers on what to do? I've tried to understand Grub 2 but always failed...
- I don't really feel like reinstalling Matlab and some other programs, can I just symlink those from my Ubuntu partition once it has a fixed mount point, and run them like usual?
1) You can chroot into your Ubuntu partition and run the "grub" program from command line: mount Ubuntu, type "chroot /path/to/ubuntu". I think you need to mount /dev, /sys and /proc before entering the chroot environment. Mount them using "mount --bind /dev /path/to/ubuntu/dev".
Once you ran grub you should type "find /boot/grub/stage1". It should list two entries: the Ubuntu one and the Fedora one. If you know which one is the Ubuntu partition (probably the first one) you type "root (hdX,X)" and then "setup (hd0)" (presuming you want to install grub in the MBR). You should probably add the Fedora entry in Ubuntu's menu.lst before running the grub shell.
2) It could work, but maybe the programs will need some libs that are not installed on Fedora. I don't know if you could actually symlink everything to make it work, it may be too much of a hassle. You could just try it out.
The thing is, it's really, really useful if all the versions where the first number is the same are mutually compatible to each other. All versions grub 1.x and all versions grub 2.x are pretty much entirely different programs, and the former can continue to be improved even as the latter matures. This worked incredibly well for apache, where apache 1.x prospered for many years after apache 2.x slowly developed into its own kind of awesome. KDE 3.x (under the project name Trinity) is still under development, even thought KDE 4.x has been mature for a very long time. Gnome 2.x will probably see continued attention while people adjust to their ... hesitation regarding Gnome 3.x.
It would be a clusterfuck if we used year.month in all those scenarios, because the newer, entirely rewritten versions are sometimes introduced in the middle of a year, and the older versions continue getting developed after their introductory year.
What would have happened if, in the middle of 2008, the new Grub came out?
We'd have Grub 2008.05 being completely incompatible with Grub 2008.06. Grub 2008.05 would continue to get upgrades and patches. What do we call the new versions? If they stay with the yyyy.mm scheme, then we get intersecting version numbers, where Grub 2009.02 is actually an improved Grub 2008.06 while Grub 2009.03 is actually an improved Grub 2008.05. Or maybe we'll call it Grub 2008.05.1, but that pretty much annihilates the point of yyyy.mm, since even though this particular Grub is mature and up to date (and intended for current usage), it is given an older version number that will confuse people. And some people might think that this newer version is just a day newer than what it's replacing.
Now, I guess you could do something like Major.Year.Month, but there are other aspects to non-dated release versioning. For example, the third (or sometimes the fourth) section, in some schemes, will explicitly tell you that we're dealing with security fixes and not feature upgrades.
A frequent updater of software might know that upgrading from killerapp 1.0.0 to killerapp 2.0.0 represents a complete rewrite of the software, upgrading from killerapp 1.0.0 to killerapp 1.1.0 represents an addition of features, and upgrading from killerapp 1.0.0 to killerapp 1.0.1 represents patching of holes and bugs. This gives you a lot more information than other schemes which only tell you "higher number means it's newer", and it makes it far less confusing when you have to decide whether or not to include a particular upgrade to your existing repository (debian stable, for instance, should generally only update the version increases that are security/bug fixes).
Edit: It doesn't seem altogether a bad idea to use a yyyy.mm versioning scheme for software suites/collections, though
I know, using year.month for the kernel or usual programs is insane. Everyone with a couple of years on linux knows the naming of them and usually the name of the programs (like ubuntu1 means it has patches by ubuntu so it's not the vanilla program).
A frequent updater of software might know that upgrading from killerapp 1.0.0 to killerapp 2.0.0 represents a complete rewrite of the software, upgrading from killerapp 1.0.0 to killerapp 1.1.0 represents an addition of features, and upgrading from killerapp 1.0.0 to killerapp 1.0.1 represents patching of holes and bugs
That is what I loved about the naming. If I sit in a system and the kernel is reported as 2.7.xx I know they are using the unstable version. If an upgrade comes and it changes from 2.6.36 to 2.6.38 I know there is gonna be changes in features and it could affect my system. 2011.03 to 2011.06 only tells me that is newer. Nothing else. Maybe is a security patch. Maybe there is new features. Maybe it's even a unstable version that slipped into the repositories!
I believe the naming is pretty good at the moment, no need to change it al all. But I guess that with 2.6 being released on 2003(2004?) it seems like the kernel is stalled. Specially after the huge change that 2.4 to 2.6 brought with al the wireless stuff and such. Even when there is huge changes behind the hood, like DRM or KMS, probably the perception in less knowledge people is that the kernel is old, or it doesn't have many new features.
I don't care though, if they want to bump it to 2.8 is well deserved with all the major introductions, but doing it out of nowhere seems kind of weird. Let's hope this isn't one of those Linus "fuck it" ideas
- I let Fedora install a new boot loader for me, but it only detected my Windows partition and not the Ubuntu one that I kept. Tomorrow I'll add the Ubuntu boot option in menu.lst, but if I want to return to the Ubuntu Grub 2-loader, can anyone give a few pointers on what to do? I've tried to understand Grub 2 but always failed...
1) You can chroot into your Ubuntu partition and run the "grub" program from command line: mount Ubuntu, type "chroot /path/to/ubuntu". I think you need to mount /dev, /sys and /proc before entering the chroot environment. Mount them using "mount --bind /dev /path/to/ubuntu/dev".
Once you ran grub you should type "find /boot/grub/stage1". It should list two entries: the Ubuntu one and the Fedora one. If you know which one is the Ubuntu partition (probably the first one) you type "root (hdX,X)" and then "setup (hd0)" (presuming you want to install grub in the MBR). You should probably add the Fedora entry in Ubuntu's menu.lst before running the grub shell.
2) It could work, but maybe the programs will need some libs that are not installed on Fedora. I don't know if you could actually symlink everything to make it work, it may be too much of a hassle. You could just try it out.
Thanks, I managed to fix it. I think I'll need to re-run update-grub in Ubuntu after every kernel update which sadly isn't optimal, how do you guys manage this when several dists use version-named kernels (ie. vmlinuz-2.6.38.6-26-etc instead of upgrading one specific like Arch does)?
Lots of repos that I set up when installing software don't have a 15 version up yet btw, right now I'm editing the repo-files and setting the release to 14 to keep YUM from swearing. The pains of early adoption I guess
Thanks, I managed to fix it. I think I'll need to re-run update-grub in Ubuntu after every kernel update which sadly isn't optimal, how do you guys manage this when several dists use version-named kernels (ie. vmlinuz-2.6.38.6-26-etc instead of upgrading one specific like Arch does)?
Lots of repos that I set up when installing software don't have a 15 version up yet btw, right now I'm editing the repo-files and setting the release to 14 to keep YUM from swearing. The pains of early adoption I guess
I had never needed to do anything special. I used to have around 3/4 different distros in an hdd and they auto did the update-grub part themselves after the installation of each new kernel.
Or you could install grub on each partition which contains each distro except for the last one so grub on the main hdx will chainload into grub in the individuals hdx1, hdx2, etc.. but that is a lousy solution.
How is the software in the final version of F15? I found the betas to be lacking in software, they were pretty barebones
can someone school me in permissions? I'm setting up an unRAID NAS and trying to run sabnzbd and sickbeard on it. For some reason I'm not able to browse full folder structures or create new folders in sickbeard, and I'm guessing its something to do with permissions.
