FC6 unstable since big yum update.

Email lists > Fedora linux discussion group

Thread: FC6 unstable since big yum update.

Search this list only Search all lists
FC6 unstable since big yum update.
United States	2007-07-15 15:18:39
I've got an FC6-64AMD setup that I hadn't done an update on in a few months. Decided to do it this weekend and, 500MB+ of packages later, I went from 2.6.20-1.2948 to 1.2962 along with all the other stuff that it comes with. I have an Nvidia display adapter that I've been using the proprietary drivers for (via livna). The upgrade causes the system to blow away my Nvidia setup for 1.2948 (guess it can't co- exist?) so I can't go backwards in grub and still get X going. This wouldn't be such a problem except my system is completely unstable, locking up after a few hours whereas before I've had 100+ day uptimes with fairly heavy usage previously. This sucks. Is there any way to undo a general 'yum update' or figure out just which out of the hundred or so packages is causing me to hang? Guess I have to me a LOT more careful when updating this system. Thought I only had that problem with Windows alas... Is there some mechanism other than a full backup/restore that I can use to protect me from an unstable update in the future? thanx for any help, -- Ben PS: Interestingly the ffmpeg-libs package insists that I need a version of faad2 < 2.5 so I have to exclude it to do an update. Dunno if that's a clue or not otherwise yum reported everything went fine.

Re: FC6 unstable since big yum update.
	2007-07-15 16:20:15

Yum should have generated a change log which will give you a place too look. Things too look at potentially is corruption in the RPM DB. If the update crashed in the middle then one of the packages was only half installed. and could cause some instability if it was an important package. rpm --force will allow you to downgrade the package to the proper NVidia rpm that was working for you. That would be a good place to start. As a work around to get your mouse cursor back in X you can go into /etc/X11/ and edit xorg.conf In the device section under NV add this line. Option "HWCusor" "Off" Save the file and restart X. That will give you your Cursor back. As for instability, one thing to check is to log in with a different user. See if ti's still unstable. Might be something in your personal settings is corrupted and causing issues. I've been using Linux for about 10 years now and unstable Linux boxes are caused by a problem that has a solution.  They rarely happen in my expience. My uptime despite being heavily used and having literally several thousand packages installed and most of them used at least occasionally is that I reboot when the power goes down or I upgrade the kernel. In the last 10 years I've rebooted a Linux machine maybe five times for "fix" a problem.  Though I am having real instability problems with Firefox and a couple apps under FC7 with an AMD 64. I suspect there's a bad RPM in the repositories right now causing some issues. The main OS runs fine but Firefox goes away anytime I look at it wrong and a few other apps have just up and crashed. The RPM DB get locked by some unknown app and I fear it's corrupted :( So the problem may be in the repositories. In my experience such problems are very short lived. As for a wipe and load, it takes a little time to reinstall all the files, especially if you load yours down like mine. But I have a seperate partition for / one for /home one for user data and a couple very large partitions for genreral data. So if I need to wipe and load it doesn't cost me anything but installed packages. Everything else stays the same. On 7/15/07, proteusguy < proteusguygmail.com">proteusguygmail.com> wrote:   I';ve got an FC6-64AMD setup that I hadn't done an update on in a few months. Decided to do it this weekend and, 500MB+ of packages later, I went from 2.6.20-1.2948 to 1.2962 along with all the other stuff that it comes with. I have an Nvidia display adapter that I've been using the proprietary drivers for (via livna). The upgrade causes the system to blow away my Nvidia setup for 1.2948 (guess it can't co- exist?) so I can't go backwards in grub and still get X going.     This wouldn';t be such a problem except my system is completely unstable, locking up after a few hours whereas before I've had 100+ day uptimes with fairly heavy usage previously. This sucks. Is there any way to undo a general 'yum update'; or figure out just which out of the hundred or so packages is causing me to hang? Guess I have to me a LOT more careful when updating this system. Thought I only had that problem with Windows alas... Is there some mechanism other than a full backup/restore that I can use to protect me from an unstable update in the future? thanx for any help,   -- Ben PS: Interestingly the ffmpeg-libs package insists that I need a version of faad2 < 2.5 so I have to exclude it to do an update. Dunno if that's a clue or not otherwise yum reported everything went fine.
Re: FC6 unstable since big yum update.
United States	2007-07-21 05:38:09
Thanx for the info Dan. I was able to get back to the previous kernel by uninstalling the new one and the machine was stable again. Something's definitely wrong in the FC6_64 kernel packages I think which is no good. The problem was getting back to the prior Nvidia driver. I don't know where yum hides its log file details but it would be nice to be able to ask it to just put this package back to the one I had before or even some version prior to the one I have. Problem is you have to know the EXACT and full name of a package to specify a different version of one you've already installed or to avoid getting the latest and all prerequisites. Sounds like something a package manager should be able to do fairly easily. Anyway - I finally ran across a mirror site for FC6 and just browsed the files until I was able to find what the exact name was and got it installed. I'm back to normal now but don't dare do a general 'yum update' now. -- Ben On Jul 15, 5:20 pm, "Dan Smith" <draci...gmail.com> wrote: > Yum should have generated a change log which will give you a place too look. > > Things too look at potentially is corruption in the RPM DB. If the update > crashed in the middle then one of the packages was only half installed. and > could cause some instability if it was an important package. > > rpm --force will allow you to downgrade the package to the proper NVidia rpm > that was working for you. That would be a good place to start. > > As a work around to get your mouse cursor back in X you can go into > /etc/X11/ and edit xorg.conf > In the device section under NV add this line. > Option "HWCusor" "Off" > > Save the file and restart X. That will give you your Cursor back. > > As for instability, one thing to check is to log in with a different user. > See if ti's still unstable. Might be something in your personal settings is > corrupted and causing issues. > > I've been using Linux for about 10 years now and unstable Linux boxes are > caused by a problem that has a solution. They rarely happen in my expience. > My uptime despite being heavily used and having literally several thousand > packages installed and most of them used at least occasionally is that I > reboot when the power goes down or I upgrade the kernel. In the last 10 > years I've rebooted a Linux machine maybe five times for "fix" a problem. > Though I am having real instability problems with Firefox and a couple apps > under FC7 with an AMD 64. I suspect there's a bad RPM in the repositories > right now causing some issues. The main OS runs fine but Firefox goes away > anytime I look at it wrong and a few other apps have just up and crashed. > The RPM DB get locked by some unknown app and I fear it's corrupted :( So > the problem may be in the repositories. In my experience such problems are > very short lived. > > As for a wipe and load, it takes a little time to reinstall all the files, > especially if you load yours down like mine. But I have a seperate partition > for / one for /home one for user data and a couple very large partitions for > genreral data. So if I need to wipe and load it doesn't cost me anything but > installed packages. Everything else stays the same. > > On 7/15/07, proteusguy <proteus...gmail.com> wrote: > > > > > I've got an FC6-64AMD setup that I hadn't done an update on in a > > few months. Decided to do it this weekend and, 500MB+ of packages > > later, I went from 2.6.20-1.2948 to 1.2962 along with all the other > > stuff that it comes with. I have an Nvidia display adapter that I've > > been using the proprietary drivers for (via livna). The upgrade causes > > the system to blow away my Nvidia setup for 1.2948 (guess it can't co- > > exist?) so I can't go backwards in grub and still get X going. > > > This wouldn't be such a problem except my system is completely > > unstable, locking up after a few hours whereas before I've had 100+ > > day uptimes with fairly heavy usage previously. This sucks. Is there > > any way to undo a general 'yum update' or figure out just which out of > > the hundred or so packages is causing me to hang? > > > Guess I have to me a LOT more careful when updating this system. > > Thought I only had that problem with Windows alas... Is there some > > mechanism other than a full backup/restore that I can use to protect > > me from an unstable update in the future? > > > thanx for any help, > > > -- Ben > > > PS: Interestingly the ffmpeg-libs package insists that I need a > > version of faad2 < 2.5 so I have to exclude it to do an update. Dunno > > if that's a clue or not otherwise yum reported everything went fine.

Re: FC6 unstable since big yum update.
	2007-07-21 11:48:14

No problem... To protect certain packages, I often protect kernel packages for example you can install a plugin for yum that allows you to lock them. ; If your using KDE I suggest Kyum or Yum-Extender. You can install yum-extender from the add packages menu. For Gnome I suggest gnome-yum if you don't have KDE libs installed. Then just scroll down to yum and you'll find some really useful yum tools. You'll also have a gui that sits atop the change log though you'll want to manually rotate that log with each start of yum. What you can do to insert that is At the beggining of the Yum script insert your favorite logrotate util or write your own. Something like yumchangelogtodaysdate.log and save it off to a new directory. The script should be in /usr/bin.  In /var/log I have yum.log rotated. So I have yum.log yum.log.1 and so on. This is default behavior on FC7 but I don't FC6 did this. I remember having to set that up. For recent changes just use the GUIs to view the change logs. #!/usr/bin/python >>> add your log rotate command here before yum does anything else. import sys try:   import yum except ImportError: There are a couple other ways. You can do a rpm -qa > filename of your choice as a cron script. That gives you a complete listing of your RPM database. A diff util can then sniff out changes between that and newer runs and create a custom change log. Also a good idea is to backup /var/lib/rpm before or after big updates. If the rpm DB gets corrupted it sometimes takes a long time to get it right. rpm -rebuildb is not the most reliable command I've run across. I use it as a last resort, though I've never had too do a wipe and load because of a corrupted rpm database. It just took several tries a time or too to get a stable rpm db after corruption mangled it. I agree that version control is sadly lacking and I also am dismayed that kernel rpms do not back up the old kernel when run. With RH 8 I had a nasty surprise. The SCSI drivers for the raid array I was using which were in RH 7 were not in RH 8 for some reason or were not detected. So when I upgraded I was unable too boot as the server had no IDE drives too boot from. Nothing but a kernel panic as a clue as too what happened. Had to hand compile a new kernel and install it off a rescue boot. RH 8 also wiped out some of the desktops in the network as some drivers did not take or were not present. Which again required manual intervention because these people had the redhat update daemon enabled. The bugs were quickly fixed but caused all sorts of havoc for a couple days. It';d be nice to have much better control over being able to force downgrades and such. There is a yum plugin but no GUI support for it. So you have to downgrade manually after installing the plugin. Either that or skip yum and manually remove and or force install downgraded packages which leaves you with potential dependancy issues that might limit your ability to upgrade other packages. On 7/21/07, proteusguy < proteusguygmail.com">proteusguygmail.com> wrote:   Thanx for the info Dan. I was able to get back to the previous kernel by uninstalling the new one and the machine was stable again. Something's definitely wrong in the FC6_64 kernel packages I think which is no good. The problem was getting back to the prior Nvidia driver. I don't know where yum hides its log file details but it would be nice to be able to ask it to just put this package back to the one I had before or even some version prior to the one I have. Problem is you have to know the EXACT and full name of a package to specify a different version of one you've already installed or to avoid getting the latest and all prerequisites. Sounds like something a package manager should be able to do fairly easily. Anyway - I finally ran across a mirror site for FC6 and just browsed the files until I was able to find what the exact name was and got it installed. I'm back to normal now but don't dare do a general 'yum update'; now.   -- Ben On Jul 15, 5:20 pm, "Dan Smith"; < draci...gmail.com">draci...gmail.com> wrote: >; Yum should have generated a change log which will give you a place too look. > > Things too look at potentially is corruption in the RPM DB. If the update >; crashed in the middle then one of the packages was only half installed. and > could cause some instability if it was an important package. > > rpm --force will allow you to downgrade the package to the proper NVidia rpm > that was working for you. That would be a good place to start. >; > As a work around to get your mouse cursor back in X you can go into >; /etc/X11/ and edit xorg.conf > In the device section under NV add this line. > Option "HWCusor" "Off" > > Save the file and restart X. That will give you  your Cursor back. > > As for instability, one thing to check is to log in with a different user. > See if ti's still unstable. Might be something in your personal settings is > corrupted and causing issues. > > I've been using Linux for about 10 years now and unstable Linux boxes are > caused by a problem that has a solution.  They rarely happen in my expience. > My uptime despite being heavily used and having literally several thousand > packages installed and most of them used at least occasionally is that I > reboot when the power goes down or I upgrade the kernel. In the last 10 > years I've rebooted a Linux machine maybe five times for "fix" a problem. > Though I am having real instability problems with Firefox and a couple apps > under FC7 with an AMD 64. I suspect there's a bad RPM in the repositories > right now causing some issues. The main OS runs fine but Firefox goes away >; anytime I look at it wrong and a few other apps have just up and crashed. > The RPM DB get locked by some unknown app and I fear it's corrupted :( So > the problem may be in the repositories. In my experience such problems are > very short lived. >; > As for a wipe and load, it takes a little time to reinstall all the files, >; especially if you load yours down like mine. But I have a seperate partition > for / one for /home one for user data and a couple very large partitions for > genreral data. So if I need to wipe and load it doesn't cost me anything but > installed packages. Everything else stays the same. > > On 7/15/07, proteusguy < proteus...gmail.com"> proteus...gmail.com> wrote: >; > > > > ; I've got an FC6-64AMD setup that I hadn't done an update on in a > > few months. Decided to do it this weekend and, 500MB+ of packages > > later, I went from 2.6.20-1.2948 to 1.2962 along with all the other > > stuff that it comes with. I have an Nvidia display adapter that I've > > been using the proprietary drivers for (via livna). The upgrade causes > > the system to blow away my Nvidia setup for 1.2948 (guess it can't co- > > exist?) so I can't go backwards in grub and still get X going. >; > > ; This wouldn';t be such a problem except my system is completely > > unstable, locking up after a few hours whereas before I've had 100+ > > day uptimes with fairly heavy usage previously. This sucks. Is there > > any way to undo a general 'yum update'; or figure out just which out of > > the hundred or so packages is causing me to hang? > > > Guess I have to me a LOT more careful when updating this system. > > Thought I only had that problem with Windows alas... Is there some >; > mechanism other than a full backup/restore that I can use to protect > > me from an unstable update in the future? > > > thanx for any help, > > > -- Ben > > > PS: Interestingly the ffmpeg-libs package insists that I need a > > version of faad2 < 2.5 so I have to exclude it to do an update. Dunno > > if that's a clue or not otherwise yum reported everything went fine.
Re: FC6 unstable since big yum update.
	2007-07-21 11:50:06

Oh forgot. With the Yum guis you get to review what will be updated and add or remove items from the list before doing a mass update on your machine. Just click on the updates tab of any of the yum guis. On 7/21/07, proteusguy < proteusguygmail.com">proteusguygmail.com> wrote:   Thanx for the info Dan. I was able to get back to the previous kernel by uninstalling the new one and the machine was stable again. Something's definitely wrong in the FC6_64 kernel packages I think which is no good. The problem was getting back to the prior Nvidia driver. I don't know where yum hides its log file details but it would be nice to be able to ask it to just put this package back to the one I had before or even some version prior to the one I have. Problem is you have to know the EXACT and full name of a package to specify a different version of one you've already installed or to avoid getting the latest and all prerequisites. Sounds like something a package manager should be able to do fairly easily. Anyway - I finally ran across a mirror site for FC6 and just browsed the files until I was able to find what the exact name was and got it installed. I'm back to normal now but don't dare do a general 'yum update'; now.   -- Ben On Jul 15, 5:20 pm, "Dan Smith"; < draci...gmail.com">draci...gmail.com> wrote: >; Yum should have generated a change log which will give you a place too look. > > Things too look at potentially is corruption in the RPM DB. If the update >; crashed in the middle then one of the packages was only half installed. and > could cause some instability if it was an important package. > > rpm --force will allow you to downgrade the package to the proper NVidia rpm > that was working for you. That would be a good place to start. >; > As a work around to get your mouse cursor back in X you can go into >; /etc/X11/ and edit xorg.conf > In the device section under NV add this line. > Option "HWCusor" "Off" > > Save the file and restart X. That will give you  your Cursor back. > > As for instability, one thing to check is to log in with a different user. > See if ti's still unstable. Might be something in your personal settings is > corrupted and causing issues. > > I've been using Linux for about 10 years now and unstable Linux boxes are > caused by a problem that has a solution.  They rarely happen in my expience. > My uptime despite being heavily used and having literally several thousand > packages installed and most of them used at least occasionally is that I > reboot when the power goes down or I upgrade the kernel. In the last 10 > years I've rebooted a Linux machine maybe five times for "fix" a problem. > Though I am having real instability problems with Firefox and a couple apps > under FC7 with an AMD 64. I suspect there's a bad RPM in the repositories > right now causing some issues. The main OS runs fine but Firefox goes away >; anytime I look at it wrong and a few other apps have just up and crashed. > The RPM DB get locked by some unknown app and I fear it's corrupted :( So > the problem may be in the repositories. In my experience such problems are > very short lived. >; > As for a wipe and load, it takes a little time to reinstall all the files, >; especially if you load yours down like mine. But I have a seperate partition > for / one for /home one for user data and a couple very large partitions for > genreral data. So if I need to wipe and load it doesn't cost me anything but > installed packages. Everything else stays the same. > > On 7/15/07, proteusguy < proteus...gmail.com"> proteus...gmail.com> wrote: >; > > > > ; I've got an FC6-64AMD setup that I hadn't done an update on in a > > few months. Decided to do it this weekend and, 500MB+ of packages > > later, I went from 2.6.20-1.2948 to 1.2962 along with all the other > > stuff that it comes with. I have an Nvidia display adapter that I've > > been using the proprietary drivers for (via livna). The upgrade causes > > the system to blow away my Nvidia setup for 1.2948 (guess it can't co- > > exist?) so I can't go backwards in grub and still get X going. >; > > ; This wouldn';t be such a problem except my system is completely > > unstable, locking up after a few hours whereas before I've had 100+ > > day uptimes with fairly heavy usage previously. This sucks. Is there > > any way to undo a general 'yum update'; or figure out just which out of > > the hundred or so packages is causing me to hang? > > > Guess I have to me a LOT more careful when updating this system. > > Thought I only had that problem with Windows alas... Is there some >; > mechanism other than a full backup/restore that I can use to protect > > me from an unstable update in the future? > > > thanx for any help, > > > -- Ben > > > PS: Interestingly the ffmpeg-libs package insists that I need a > > version of faad2 < 2.5 so I have to exclude it to do an update. Dunno > > if that's a clue or not otherwise yum reported everything went fine.

[1-5]

about | contact Other archives ( Real Estate discussion Medical topics )