Bug #5775 is not fixed! Sad, but true. :( I did clean 0.8 beta1 install under vmware, mounted my local mirror, installed pacman-rc, but all was _exactly_ the same as reported here: http://bugs.archlinux.org/task/5775#comment12437 The _only_ difference is that -Rs worked without errors. Then I removed /var/lib/pacman/current (just to be sure that it wasn't corrupted ny pacman2 when doing first -Syu after install), but syncing repo with pacman3 still results in broken db and problem with xorg-res-utils. -- Roman Kyrylych (Роман Кирилич)
2007/2/6, Roman Kyrylych <roman.kyrylych@gmail.com>:
Bug #5775 is not fixed! Sad, but true. :(
I did clean 0.8 beta1 install under vmware, mounted my local mirror, installed pacman-rc, but all was _exactly_ the same as reported here: http://bugs.archlinux.org/task/5775#comment12437 The _only_ difference is that -Rs worked without errors.
Then I removed /var/lib/pacman/current (just to be sure that it wasn't corrupted ny pacman2 when doing first -Syu after install), but syncing repo with pacman3 still results in broken db and problem with xorg-res-utils.
The good thing is that this report: http://bugs.archlinux.org/?getfile=884 which was mentioned in the same bugreport (IIRC it was reported on IRC by Shadowhand) is fixed. -- Roman Kyrylych (Роман Кирилич)
On 2/6/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/6, Roman Kyrylych <roman.kyrylych@gmail.com>:
Bug #5775 is not fixed! Sad, but true. :(
Roman, I don't know what it is, but I just did a clean install of Arch in Qemu and could not duplicate your bug. Everything worked fine for me. I don't know why on earth this would be a one-time bug and only in your vmware installation, but I don't know what we can do about it if it isn't duplicatable. -Dan
2007/2/8, Dan McGee <dpmcgee@gmail.com>:
On 2/6/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/6, Roman Kyrylych <roman.kyrylych@gmail.com>:
Bug #5775 is not fixed! Sad, but true. :(
Roman, I don't know what it is, but I just did a clean install of Arch in Qemu and could not duplicate your bug. Everything worked fine for me. I don't know why on earth this would be a one-time bug and only in your vmware installation, but I don't know what we can do about it if it isn't duplicatable.
Can I send you my huge vmware file? ;) I will try to reproduce it again with xorg-res-utils-1.0.3-1. Please reproduce all steps as I did and then diff my debug output with your. I'ms sure there must be some differences. Also note, that I saw this bug first on real machine, and was able to reproduce it clearly. I can reproduce it in virtualbox (which is based on qemu) too. It is clearly not one-time bug. Even if it's magically fixed with new versions of something except pacman - there's no evidence that this won't happen again. :( -- Roman Kyrylych (Роман Кирилич)
On 2/8/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/8, Dan McGee <dpmcgee@gmail.com>:
On 2/6/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/6, Roman Kyrylych <roman.kyrylych@gmail.com>:
Bug #5775 is not fixed! Sad, but true. :(
Roman, I don't know what it is, but I just did a clean install of Arch in Qemu and could not duplicate your bug. Everything worked fine for me. I don't know why on earth this would be a one-time bug and only in your vmware installation, but I don't know what we can do about it if it isn't duplicatable.
Can I send you my huge vmware file? ;) I will try to reproduce it again with xorg-res-utils-1.0.3-1.
Please reproduce all steps as I did and then diff my debug output with your. I'ms sure there must be some differences.
Also note, that I saw this bug first on real machine, and was able to reproduce it clearly. I can reproduce it in virtualbox (which is based on qemu) too. It is clearly not one-time bug.
Even if it's magically fixed with new versions of something except pacman - there's no evidence that this won't happen again. :(
Which version of pacman are you trying this with? Dan, cactus, and myself all tried it with the most recent version. And got nothing.
2007/2/8, Aaron Griffin <aaronmgriffin@gmail.com>:
On 2/8/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/8, Dan McGee <dpmcgee@gmail.com>:
On 2/6/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/6, Roman Kyrylych <roman.kyrylych@gmail.com>:
Bug #5775 is not fixed! Sad, but true. :(
Roman, I don't know what it is, but I just did a clean install of Arch in Qemu and could not duplicate your bug. Everything worked fine for me. I don't know why on earth this would be a one-time bug and only in your vmware installation, but I don't know what we can do about it if it isn't duplicatable.
Can I send you my huge vmware file? ;) I will try to reproduce it again with xorg-res-utils-1.0.3-1.
Please reproduce all steps as I did and then diff my debug output with your. I'ms sure there must be some differences.
Also note, that I saw this bug first on real machine, and was able to reproduce it clearly. I can reproduce it in virtualbox (which is based on qemu) too. It is clearly not one-time bug.
Even if it's magically fixed with new versions of something except pacman - there's no evidence that this won't happen again. :(
Which version of pacman are you trying this with? Dan, cactus, and myself all tried it with the most recent version. And got nothing.
My last debug report was done with rc9. I can reproduce it with any version. Last time I reproduced it on a freshly installed 0.8beta1, synced Current from ~10 days old full local mirror and repeated all steps as in my last debug report. Are you sure you did pacman -S _exactly_ as in my report and with clean Current db? Anyway please diff your debug output with mine. If you want I can reproduce the same steps on new real machine at my work. -- Roman Kyrylych (Роман Кирилич)
My last debug report was done with rc9. I can reproduce it with any version. Last time I reproduced it on a freshly installed 0.8beta1, synced Current from ~10 days old full local mirror and repeated all steps as in my last debug report. Are you sure you did pacman -S _exactly_ as in my report and with clean Current db? Anyway please diff your debug output with mine. If you want I can reproduce the same steps on new real machine at my work.
The bug report was a bit hard to follow. I *think* I followed your steps, but did not get the same results. Could you possibly reply with the exact commands you used, step by step, in a very simple layout? like.. 1. pacman -S foox blah foo 2. pacman -Rs foox blah foo 3. pacman -S foo <note the breakage> 4. profit?
2007/2/8, eliott@cactuswax.net <eliott@cactuswax.net>:
My last debug report was done with rc9. I can reproduce it with any version. Last time I reproduced it on a freshly installed 0.8beta1, synced Current from ~10 days old full local mirror and repeated all steps as in my last debug report. Are you sure you did pacman -S _exactly_ as in my report and with clean Current db? Anyway please diff your debug output with mine. If you want I can reproduce the same steps on new real machine at my work.
The bug report was a bit hard to follow. I *think* I followed your steps, but did not get the same results.
Could you possibly reply with the exact commands you used, step by step, in a very simple layout?
like..
1. pacman -S foox blah foo 2. pacman -Rs foox blah foo 3. pacman -S foo <note the breakage> 4. profit?
I used these commands to get info: rm -rf /var/lib/pacman/current pacman3 -Sy yes | pacman3 -S xorg-server xorg-clients xf86-input-keyboard xf86-input-mouse xf86-video-vmware --debug > S-debug.txt pacman3 -Qi xorg-res-utils > Qi.txt cat /var/lib/pacman/current/xorg-res-utils-1.0.0-1/depends > current-xorg-res-utils-depends.txt cat /var/lib/pacman/local/xorg-res-utils-1.0.0-1/depends > local-xorg-res-utils-depends.txt yes | pacman3 -Rs xorg-server xorg-clients xf86-input-keyboard xf86-input-mouse xf86-video-vmware --debug > Rs-debug.txt Same for --debug=3, see S-debug3.txt and Rs-debug3.txt For output see http://bugs.archlinux.org/task/5775#comment12437 (except pacman3 doesn't show "warning: cannot find package "mesa" or anything that provides it!" 75 times) I think I know why you cannot reproduce this. When you did pacman3 -S ... you got xorg-res-utils-1.0.3-1, right? I updated one of my machines today and db corruption has gone. From the time #5775 was opened until recent xorg-res-utils update it was valid. I have 2 snapshots of full ftp.archlinux.org mirror done with rsync. Older (~10 days ago) has xorg-res-utils-1.0.0-1, newer was done this night and has 1.0.3-1. I will try both today and make a diff of output. I think we must find a reason of db corruption now, so this won't happen again in future. -- Roman Kyrylych (Роман Кирилич)
On 2/8/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
I think I know why you cannot reproduce this. When you did pacman3 -S ... you got xorg-res-utils-1.0.3-1, right? I updated one of my machines today and db corruption has gone. From the time #5775 was opened until recent xorg-res-utils update it was valid. I have 2 snapshots of full ftp.archlinux.org mirror done with rsync. Older (~10 days ago) has xorg-res-utils-1.0.0-1, newer was done this night and has 1.0.3-1. I will try both today and make a diff of output. I think we must find a reason of db corruption now, so this won't happen again in future.
Just to be sure: there was an RC a while back that corrupted the DB. This install has NOT been touched with that RC, has it? It is completely, 100%, new?
2007/2/8, Aaron Griffin <aaronmgriffin@gmail.com>:
On 2/8/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
I think I know why you cannot reproduce this. When you did pacman3 -S ... you got xorg-res-utils-1.0.3-1, right? I updated one of my machines today and db corruption has gone. From the time #5775 was opened until recent xorg-res-utils update it was valid. I have 2 snapshots of full ftp.archlinux.org mirror done with rsync. Older (~10 days ago) has xorg-res-utils-1.0.0-1, newer was done this night and has 1.0.3-1. I will try both today and make a diff of output. I think we must find a reason of db corruption now, so this won't happen again in future.
Just to be sure: there was an RC a while back that corrupted the DB. This install has NOT been touched with that RC, has it? It is completely, 100%, new?
No, of course. I did all tests on clean system and reinstalled it after each test. I will check my guess about xorg-res-utils version after ~2 hours. -- Roman Kyrylych (Роман Кирилич)
I found the reason! The problem is that xorg-res-utils-1.0.0-1.pkg.tar.gz contained wrong .PKGINFO all time until 1.0.3-1 update! # Generated by makepkg 2.9.8 # Sat Feb 18 10:39:05 UTC 2006 pkgname = xorg-res-utils pkgver = 1.0.0-1 pkgdesc = X.Org X11 resource utilities url = http://xorg.freedesktop.org/ builddate = Sat Feb 18 10:39:05 2006 packager = Arch Linux (http://www.archlinux.org) size = 146504 arch = i686 depend = mesa depend = libxxf86dga depend = libxi depend = libxxf86misc depend = libxtst depend = libxinerama depend = libxv depend = libdmx depend = libxft depend = libxaw While PKGBUILD always contained depends=(libxaw) and thus gensync generated the following xorg-res-utils-1.0.0-1/depends in our Current repo: %DEPENDS% libxaw When pacman (any version) installed xorg-res-utils-1.0.0-1 it generated the following /var/lib/pacman/local/xorg-res-utils-1.0.0-1/depends: %DEPENDS% mesa libxxf86dga libxi libxxf86misc libxtst libxinerama libxv libdmx libxft libxaw %REQUIREDBY% xorg-clients So, pacman and gensync did everything correctly. I think .PKGINFO and current.db mismatch was due to PKGBUILD's depends array was modified on dev's machine after building a package, but new package wasn't rebuilt. -- Roman Kyrylych (Роман Кирилич)
On 2/9/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
I found the reason! The problem is that xorg-res-utils-1.0.0-1.pkg.tar.gz contained wrong .PKGINFO all time until 1.0.3-1 update!
Hooray! Looks like we only have 2 (and a half, if you count documentation) bugs remaining. http://bugs.archlinux.org/task/6316 We're behind the release schedule by a few days, but that is to be expected. I'm going to run through some of the output issues today/tonight and push one last RC (with fixed man pages this time)... expect it in a few hours or so
2007/2/9, Aaron Griffin <aaronmgriffin@gmail.com>:
On 2/9/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
I found the reason! The problem is that xorg-res-utils-1.0.0-1.pkg.tar.gz contained wrong .PKGINFO all time until 1.0.3-1 update!
Hooray! Looks like we only have 2 (and a half, if you count documentation) bugs remaining.
http://bugs.archlinux.org/task/6316
We're behind the release schedule by a few days, but that is to be expected.
I'm going to run through some of the output issues today/tonight and push one last RC (with fixed man pages this time)... expect it in a few hours or so
What about #4744? -- Roman Kyrylych (Роман Кирилич)
#4744 seems to be added to the list already. Though, i don't know how to test / verify this. I don't know this error using pacman3, have XFS and never got in struggle with pacman. I can't think of a common error, since it would appear more often. Anyone ideas for a starting point? Kind regards, Georg On 2/9/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/2/9, Aaron Griffin <aaronmgriffin@gmail.com>:
On 2/9/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
I found the reason! The problem is that xorg-res-utils-1.0.0-1.pkg.tar.gz contained wrong .PKGINFO all time until 1.0.3-1 update!
Hooray! Looks like we only have 2 (and a half, if you count documentation) bugs remaining.
http://bugs.archlinux.org/task/6316
We're behind the release schedule by a few days, but that is to be expected.
I'm going to run through some of the output issues today/tonight and push one last RC (with fixed man pages this time)... expect it in a few hours or so
What about #4744?
-- Roman Kyrylych (Роман Кирилич) _______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://www.archlinux.org/mailman/listinfo/pacman-dev
On 2/9/07, Georg Grabler <ggrabler@gmail.com> wrote:
#4744 seems to be added to the list already.
Though, i don't know how to test / verify this. I don't know this error using pacman3, have XFS and never got in struggle with pacman. I can't think of a common error, since it would appear more often.
Anyone ideas for a starting point?
Well, I looked at it, and the forum threads. The odd thing is that pacman didn't bail with the md5sum checking there, meaning it actually thought the package was OK. I honestly think this means that the package itself, when added to the repo, was corrupted.
Hmh, i don't know how that exactly works, i never have been TU or able to commti any package. The MD5SUMs seem to be saved server side in the package caches. This provides two points of failrue: - The packge wasn't completely transferred. - The package was already corrupted before uploading. The 2nd point should be checked by the packagers, installing the packages. I still think that the md5sums should actually be provided (how ever that could be implemented) by the maintainers, and just be taken to the repo by your server side scripts (?). In this case, the package would never appear if the sums don't match. Though, what is a bit curious: They said after rebooting the system it worked. That would mean the package was not corrupted. Or was this just on the one bug report i read? Kind regards, Georg On 2/9/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 2/9/07, Georg Grabler <ggrabler@gmail.com> wrote:
#4744 seems to be added to the list already.
Though, i don't know how to test / verify this. I don't know this error using pacman3, have XFS and never got in struggle with pacman. I can't think of a common error, since it would appear more often.
Anyone ideas for a starting point?
Well, I looked at it, and the forum threads. The odd thing is that pacman didn't bail with the md5sum checking there, meaning it actually thought the package was OK. I honestly think this means that the package itself, when added to the repo, was corrupted.
_______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://www.archlinux.org/mailman/listinfo/pacman-dev
participants (5)
-
Aaron Griffin
-
Dan McGee
-
eliott＠cactuswax.net
-
Georg Grabler
-
Roman Kyrylych