Hello list.
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Regards.
Dominik Składanowski wrote:
Hello list.
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Regards.
Have you reported the issue to the RPMForge lists ?
here is a link : http://lists.rpmforge.net/mailman/listinfo/users
- K
Karanbir Singh wrote:
Dominik Składanowski wrote:
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Regards.
Have you reported the issue to the RPMForge lists ?
here is a link : http://lists.rpmforge.net/mailman/listinfo/users
See http://lists.rpmforge.net/pipermail/users/2005-November/000110.html (Dominik already has answers from the users list, but this might clarify it for others also).
Ralph
On Mon, Dec 05, 2005 at 12:05:01AM +0100, Ralph Angenendt wrote:
See http://lists.rpmforge.net/pipermail/users/2005-November/000110.html (Dominik already has answers from the users list, but this might clarify it for others also).
Hi Ralph,
This suggests that that list has had postings last days. Are the archives by any change delayed? (I'm asking 'cause maybe here also 'the admins may not know about it')
Regards,
Henk van Lingen wrote:
On Mon, Dec 05, 2005 at 12:05:01AM +0100, Ralph Angenendt wrote:
See http://lists.rpmforge.net/pipermail/users/2005-November/000110.html (Dominik already has answers from the users list, but this might clarify it for others also).
Hi Ralph,
This suggests that that list has had postings last days. Are the archives by any change delayed? (I'm asking 'cause maybe here also 'the admins may not know about it')
Nope, no answer to that yet. But yes, the list has postings - and so do the archives:
http://lists.rpmforge.net/pipermail/users/2005-December/thread.html has the mails from December.
Ralph
Dominik Składanowski wrote:
Hello list.
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Regards.
Within the last minute, I successfully updated libdvdread from Dag's site. Here's my /etc/yum.repos.d/Dag.repo ------- [dag] name=Dag RPM Repository for Red Hat Enterprise Linux baseurl=http://www.mirrorservice.org/sites/apt.sw.be/redhat/el$releasever/en/i386/da... failovermethod=priority enabled=1 gpgkey=http://dag.wieers.com/packages/RPM-GPG-KEY.dag.txt gpgcheck=1
------
Change that to "within the last 5 minutes".
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Within the last minute, I successfully updated libdvdread from Dag's site. Here's my /etc/yum.repos.d/Dag.repo
[dag] name=Dag RPM Repository for Red Hat Enterprise Linux baseurl=http://www.mirrorservice.org/sites/apt.sw.be/redhat/el$releasever/en/i386/da...
failovermethod=priority enabled=1 gpgkey=http://dag.wieers.com/packages/RPM-GPG-KEY.dag.txt gpgcheck=1
Change that to "within the last 5 minutes".
I tried to use your dag.repo base URL, but still the same. The problem is perl-MIME-tools-5.418-1.2.el4.rf.noarch.rpm package.
Regards.
On Sun, Dec 04, 2005 at 07:48:21PM +0100, Dominik Składanowski wrote:
I tried to use your dag.repo base URL, but still the same. The problem is perl-MIME-tools-5.418-1.2.el4.rf.noarch.rpm package.
Well, last week I had this prob:
Downloading Packages: (1/7): perl-IO-Socket-SSL 100% |=========================| 35 kB 00:00 http://archive.cs.uu.nl/mirror/dag.wieers/redhat/el4/en/x86_64/dag/RPMS/perl...: [Errno -1] Package does not match checksum Trying other mirror. (2/7): perl-Authen-SASL-2 100% |=========================| 35 kB 00:00 (3/7): perl-Convert-ASN1- 100% |=========================| 33 kB 00:00 (4/7): perl-Digest-SHA1-2 100% |=========================| 19 kB 00:00 (5/7): perl-Net-LDAP-0.32 100% |=========================| 304 kB 00:00 (6/7): perl-Net-SSLeay-1. 100% |=========================| 199 kB 00:00 (7/7): perl-Digest-HMAC-1 100% |=========================| 11 kB 00:00
Error Downloading Packages: perl-IO-Socket-SSL - 0.97-1.2.el4.rf.noarch: failure: RPMS/perl-IO-Socket-SSL-0.97-1.2.el4.rf.noarch.rpm from dag: [Errno 256] No more mirrors to try.
As I'm mirroring DAG (http://archive.cs.uu.nl/mirror/dag.wieers/), I tried to sync the stuff. I couldn't connect to apt.sw.be for several days. Now I can, and just refreshed my mirror. The problem remains. There seems to be wrong RPM's sometimes.
I mailed Dag last week about it, but like before this year (iirc): no answer. He seems to be very busy or very away, and this DAG stuff seems to me a bit of a one-man show and thus a bit flaky at times? Too bad, as it is a great project.
Cheers,
On 12/4/05, Henk van Lingen henkvl@cs.uu.nl wrote:
I mailed Dag last week about it, but like before this year (iirc): no answer. He seems to be very busy or very away, and this DAG stuff seems to me a bit of a one-man show and thus a bit flaky at times? Too bad, as it is a great project.
You considered this, I assume:
http://lists.centos.org/pipermail/centos/2005-November/014553.html
Greg
On Sun, Dec 04, 2005 at 01:05:38PM -0700, Greg Knaddison wrote:
I mailed Dag last week about it, but like before this year (iirc): no answer. He seems to be very busy or very away, and this DAG stuff seems to me a bit of a one-man show and thus a bit flaky at times? Too bad, as it is a great project.
You considered this, I assume:
http://lists.centos.org/pipermail/centos/2005-November/014553.html
What to consider about it? If on the main dag site/mirror the admins "don't know something has been wrong for the last couple of weeks", the dag-project is indeed a bit flaky at times, right?
I looked around for the state of affairs on dag packages and/or mirrors. For example in the archive of the just mentioned RPMforge. Not a single message in december yet.
Cheers,
On 12/4/05, Henk van Lingen henkvl@cs.uu.nl wrote:
What to consider about it? If on the main dag site/mirror the admins "don't know something has been wrong for the last couple of weeks", the dag-project is indeed a bit flaky at times, right?
I looked around for the state of affairs on dag packages and/or mirrors. For example in the archive of the just mentioned RPMforge. Not a single message in december yet.
What to consider is that you (more dominik) are biting a hand that feeds you for free not only in giving you rebuilt rpms that magically work on your system but also that the site is a test bed for one of the most important pieces of software in the world used by millions of people a day (Apache).
So maybe a little patience, assistance in solving the problem, and perhaps even gratitude/donation would be in order rather than just "OMG, dag's site provided me an RPM and then the update is broken and I emailed him and it has taken a week to solve the problem!" which is how I read you and Dominik's messages.
Greg
On Sun, Dec 04, 2005 at 01:28:09PM -0700, Greg Knaddison wrote:
What to consider is that you (more dominik) are biting a hand that feeds you for free not only in giving you rebuilt rpms that magically work on your system but also that the site is a test bed for one of the most important pieces of software in the world used by millions of people a day (Apache).
So maybe a little patience, assistance in solving the problem, and perhaps even gratitude/donation would be in order rather than just "OMG, dag's site provided me an RPM and then the update is broken and I emailed him and it has taken a week to solve the problem!" which is how I read you and Dominik's messages.
I'm not biting anyone. I'm very happy with the stuff Dag does. I'm just saying it isn't very reliabe and informative at times. Concerning the donation: I'm mirroring his stuff (except fedora/ at the moment) for some years now. Last year I mailed him about it, in case he wanted mirrors used (no answer), I saw this week my mirror is now included (without me knowing it) on his website. That's fine with me. (I have an apache mirror on the same box, btw). For a system as popular as Dag's, it doesn't look sane to me to have your main mirror on a testbox, right? There have to be other mirrors/ people who would be happy to help him, i think.
Maybe that RPMforge-thing is a platform to discus things like this. But for all I can see it is very quiet there. Should I have mailed them instead of Dag about broken RPM's? Maybe, I don't know.
Regards,
On Sun, 4 Dec 2005, Henk van Lingen wrote:
On Sun, Dec 04, 2005 at 01:28:09PM -0700, Greg Knaddison wrote:
What to consider is that you (more dominik) are biting a hand that feeds you for free not only in giving you rebuilt rpms that magically work on your system but also that the site is a test bed for one of the most important pieces of software in the world used by millions of people a day (Apache).
So maybe a little patience, assistance in solving the problem, and perhaps even gratitude/donation would be in order rather than just "OMG, dag's site provided me an RPM and then the update is broken and I emailed him and it has taken a week to solve the problem!" which is how I read you and Dominik's messages.
I'm not biting anyone. I'm very happy with the stuff Dag does. I'm just saying it isn't very reliabe and informative at times. Concerning the donation: I'm mirroring his stuff (except fedora/ at the moment) for some years now. Last year I mailed him about it, in case he wanted mirrors used (no answer), I saw this week my mirror is now included (without me knowing it) on his website. That's fine with me. (I have an apache mirror on the same box, btw). For a system as popular as Dag's, it doesn't look sane to me to have your main mirror on a testbox, right? There have to be other mirrors/ people who would be happy to help him, i think.
Maybe that RPMforge-thing is a platform to discus things like this. But for all I can see it is very quiet there. Should I have mailed them instead of Dag about broken RPM's? Maybe, I don't know.
Guys, I don't have the time to run after HEAnet to fix things. Most of the times, after a while, everything is back ok. Sometimes Matthias (freshrpms) can remove files from the mirror so rsync can re-sync broken files. (But that was when HEAnet had NFS corruption problems) Matthias is the only person that can access my files since it's being hosted under his documentroot at HEAnet.
I requested access 2 times, but did not get a reply. I bet they have the same email-problem as me :-)
Other problems are due to the experimental Apache they are using (since they are one of the largest mirrors on this planet it's important for the Apache project) or other technical problems. (NFS, rsync) Since I don't have access to the mirror, it's very hard to debug the problem I cannot verify eg. that a file is wrong on disk or just goes hay-wire when being send by Apache. If mirrors do not have this problem, it's not an Apache problem, but since all mirrors mirror from HEAnet, it could still be an NFS corruption problem or something else.
The only way for me to fix inidividual reported files is by bumping the release number, rebuilding and uploading to my server.
Since I don't know how much traffic is being used every day it's impossible for me to consider other mirrors as main mirror.
I get swamped by mails of people (about 50 per day) and I have no spare time anymore. If people would report to the RPMforge mailinglists, most of the duplicate reports would be gone.
If you have problems, the best way to get my attention (by email) is to send an analysis of the problem. E.g. if you only say this or that file seems to be corrupt, I do not have the time to investigate myself.
I have had harddisk problems (several times this year), since last week my buildsystem is crashing randomly (after 15min to 1 hours of running) and I do not even have the time to look into that. (I spend my whole sunday analysing and swapping hardware, to no avail) So that makes it impossible to log on to my system, build something else. And since I'm seldom home I can't even reboot the damn thing.
RPMforge currently only comprises of Matthias, Dries and me. Other people are welcome to send fixes or problem-analysis. If you want to help, please don't simply send a report. If you have suggestions on how we can provide more information so you can help analysing problems, let me know.
We currently provide buildlogs, SPEC files, patches, SVN commitlogs and packagelists and buildenv packagelists.
But besides all that, I'm doing fine.
Kind regards, -- dag wieers, dag@wieers.com, http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power]
On Mon, Dec 05, 2005 at 03:04:08PM +0100, Dag Wieers wrote:
For a system as popular as Dag's, it doesn't look sane to me to have your main mirror on a testbox, right? There have to be other mirrors/ people who would be happy to help him, i think.
Since I don't know how much traffic is being used every day it's impossible for me to consider other mirrors as main mirror.
Hi Dag,
Could you elaborate on this? It could very well be that I'm naive (how much data are we talking?), but wouldn't it be possible to have a box somewhere to which only you/rpmforge and the mirrors can connect and let the mirrors sync from this one? If I were to supply you such a box, (just to improve the quality of my mirror :-)) what should the specs be?
I get swamped by mails of people (about 50 per day) and I have no spare time anymore. If people would report to the RPMforge mailinglists, most of the duplicate reports would be gone.
I just subscribed to the users list :-)
RPMforge currently only comprises of Matthias, Dries and me. Other people are welcome to send fixes or problem-analysis. If you want to help, please don't simply send a report. If you have suggestions on how we can provide more information so you can help analysing problems, let me know.
Maybe the RPMforge-homepage can announce known and/or current probs in hardware and/or packages?
But besides all that, I'm doing fine.
Good to hear :-)
Regards,
On Mon, 5 Dec 2005, Henk van Lingen wrote:
On Mon, Dec 05, 2005 at 03:04:08PM +0100, Dag Wieers wrote:
For a system as popular as Dag's, it doesn't look sane to me to have your main mirror on a testbox, right? There have to be other mirrors/ people who would be happy to help him, i think.
Since I don't know how much traffic is being used every day it's impossible for me to consider other mirrors as main mirror.
Could you elaborate on this? It could very well be that I'm naive (how much data are we talking?), but wouldn't it be possible to have a box somewhere to which only you/rpmforge and the mirrors can connect and let the mirrors sync from this one? If I were to supply you such a box, (just to improve the quality of my mirror :-)) what should the specs be?
The repository is now about 25GB data and this only increases over time. Best bet is to have 5GB additional every year.
I currently have 1 box that only HEAnet can connect to. This used to be the main server until I got 20GB traffic per day (almost 3 years ago). At that point it costed me too much personally and first Axel (atrpms) and then Matthias (freshrpms) offered me to host it.
But I have no clue how much bandwidth is used every day. HEAnet offers rsync access as well for all mirrors (and individuals), so I can only assume this is a considerable amount that far exceeds 20GB/day.
RPMforge currently only comprises of Matthias, Dries and me. Other people are welcome to send fixes or problem-analysis. If you want to help, please don't simply send a report. If you have suggestions on how we can provide more information so you can help analysing problems, let me know.
Maybe the RPMforge-homepage can announce known and/or current probs in hardware and/or packages?
Well, most of these problems affect only 1 distribution or only 1 builder. The right place to discuss it should be the users mailinglist right now, the homepage is not actively maintained. (I wish it was though)
I yesterday rebuild all metadata again and I removed the createrepo cache. Can one report on the users mailinglist if they still have problems with corrupt checksums ?
If this fixes it, it's a createrepo bug.
Thanks.
Kind regards, -- dag wieers, dag@wieers.com, http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power]
Dag Wieers wrote:
I yesterday rebuild all metadata again and I removed the createrepo cache. Can one report on the users mailinglist if they still have problems with corrupt checksums ?
If this fixes it, it's a createrepo bug.
Yes, this fixed it. There aren't any more broken checksums on the mirror at the moment.
Can you file that in yum's bugzilla? If you don't have the time for it: How can I recreate that bug on a local repository?
Ah yes, and thanks for fixing it ...
Ralph
Ralph Angenendt wrote:
Dag Wieers wrote:
I yesterday rebuild all metadata again and I removed the createrepo cache. Can one report on the users mailinglist if they still have problems with corrupt checksums ?
If this fixes it, it's a createrepo bug.
Yes, this fixed it. There aren't any more broken checksums on the mirror at the moment.
Can you file that in yum's bugzilla? If you don't have the time for it: How can I recreate that bug on a local repository?
the new createrepo 0.4.3 introduces a new 'feature' that it will cache previous metadata for pkgs that have not changed - so that it saves time rebuilding the md for a repo when you have a large number of pkgs, but only a few have changed.
to recreate :
use createrepo-0.4.3 and "createrepo <dir to repo>" proceed to change a few pkgs and rerun.
What I am guessing is that something on those pkgs is changing eg. a resign or something - that causes the checksum to change - but does not hit the radar of createrepo - perhaps createrepo is only checking for a rpm param like buid time or something... not sure, worth a peek in the code to verify. so while the checksum has changed, createrepo does not regen md for that package.
Karanbir Singh wrote:
Ralph Angenendt wrote:
Can you file that in yum's bugzilla? If you don't have the time for it: How can I recreate that bug on a local repository?
to recreate :
use createrepo-0.4.3 and "createrepo <dir to repo>"
createrepo -c /path/to/cache repo/
proceed to change a few pkgs and rerun.
Yes. Only seems to happen, when checksum changes occur by signing the package. Just rebuilding the package with a different release or version doesn't trigger the change.
I'm gonna file that.
Ralph
Ralph Angenendt wrote:
Yes. Only seems to happen, when checksum changes occur by signing the package. Just rebuilding the package with a different release or version doesn't trigger the change.
I'm gonna file that.
FYI: http://devel.linux.duke.edu/bugzilla/show_bug.cgi?id=530
Ralph
On Wed, 7 Dec 2005, Ralph Angenendt wrote:
Karanbir Singh wrote:
Ralph Angenendt wrote:
Can you file that in yum's bugzilla? If you don't have the time for it: How can I recreate that bug on a local repository?
to recreate :
use createrepo-0.4.3 and "createrepo <dir to repo>"
createrepo -c /path/to/cache repo/
proceed to change a few pkgs and rerun.
Yes. Only seems to happen, when checksum changes occur by signing the package. Just rebuilding the package with a different release or version doesn't trigger the change.
I'm gonna file that.
This is very likely. Thanks for analysing and verifying the problem !
Kind regards, -- dag wieers, dag@wieers.com, http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power]
I yesterday rebuild all metadata again and I removed the createrepo cache. Can one report on the users mailinglist if they still have problems with corrupt checksums ?
If this fixes it, it's a createrepo bug.
As far as I can see everything is OK now, both dag and dries pass through your test after a resync a few hours back (as compared to yesterday when both failed on roughly the same files).
Cheers, MaZe.
Quoting Maciej ¯enczykowski maze@cela.pl:
I yesterday rebuild all metadata again and I removed the createrepo
cache.
Can one report on the users mailinglist if they still have
problems
with corrupt checksums ?
If this fixes it, it's a createrepo bug.
As far as I can see everything is OK now, both dag and dries pass through your test after a resync a few hours back (as compared to yesterday when both failed on roughly the same files).
I haven't done an rsync yesterday, you can check the date of the last apt/oldyum/newyum generation in the file http://apt.sw.be/dries/TIMESTAMP So.. if something failed yesterday, it should still fail :)
kind regards, Dries
re: Dag's repo etc...
would someone please consider hosting Dag's repo(s) on 24/7 access production servers and giving him and others access please?
it appears to me that so many people rely on Dag's "stuff" that it should not be hosted on an experimental visioned platform regardless of the goodwill intentions from the upstream etc...
Dag, i would do it yet i just do not have "some" of the extra resources right now... sorry...
- rh
-- Robert Hanson - Abba Communications Computer & Internet Services (509) 624-7159 - www.abbacomm.net
What to consider about it? If on the main dag site/mirror the admins "don't know something has been wrong for the last couple of weeks", the dag-project is indeed a bit flaky at times, right?
I looked around for the state of affairs on dag packages and/or mirrors. For example in the archive of the just mentioned RPMforge. Not a single message in december yet.
What to consider is that you (more dominik) are biting a hand that feeds you for free not only in giving you rebuilt rpms that magically work on your system but also that the site is a test bed for one of the most important pieces of software in the world used by millions of people a day (Apache).
So maybe a little patience, assistance in solving the problem, and perhaps even gratitude/donation would be in order rather than just "OMG, dag's site provided me an RPM and then the update is broken and I emailed him and it has taken a week to solve the problem!" which is how I read you and Dominik's messages.
I don't bit a hand because of wrong checksum. I understand Dag repo is Dag's good will and patience is recomended. I am only confused about no one can answer what's going on.
Regards.
On Sunday 04 December 2005 19:48, Dominik Składanowski wrote:
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Within the last minute, I successfully updated libdvdread from Dag's site. Here's my /etc/yum.repos.d/Dag.repo
[dag] name=Dag RPM Repository for Red Hat Enterprise Linux baseurl=http://www.mirrorservice.org/sites/apt.sw.be/redhat/el$releasever /en/i386/dag/
failovermethod=priority enabled=1 gpgkey=http://dag.wieers.com/packages/RPM-GPG-KEY.dag.txt gpgcheck=1
Change that to "within the last 5 minutes".
I tried to use your dag.repo base URL, but still the same. The problem is perl-MIME-tools-5.418-1.2.el4.rf.noarch.rpm package.
Temporary fix: download and install the package for el4 from http://dries.ulyssis.org/rpm/packages/perl-MIME-tools/info.html I've changed the release in the subversion so the bad package gets replaced.
kind regards, Dries
Continuing previous topic: http://lists.centos.org/pipermail/centos/2005-November/015586.html, I would like to ask: anybody has the same problem? Still I can't update this few packages.
Within the last minute, I successfully updated libdvdread from Dag's site. Here's my /etc/yum.repos.d/Dag.repo
[dag] name=Dag RPM Repository for Red Hat Enterprise Linux baseurl=http://www.mirrorservice.org/sites/apt.sw.be/redhat/el$releasever /en/i386/dag/
failovermethod=priority enabled=1 gpgkey=http://dag.wieers.com/packages/RPM-GPG-KEY.dag.txt gpgcheck=1
Change that to "within the last 5 minutes".
I tried to use your dag.repo base URL, but still the same. The problem is perl-MIME-tools-5.418-1.2.el4.rf.noarch.rpm package.
Temporary fix: download and install the package for el4 from http://dries.ulyssis.org/rpm/packages/perl-MIME-tools/info.html I've changed the release in the subversion so the bad package gets replaced.
It works. Thanks Dries.
Best regards.
-
Dag Wieers -
Dominik Składanowski -
dries -
Dries Verachtert
-
Greg Knaddison -
Henk van Lingen -
Karanbir Singh -
Maciej Żenczykowski -
Ralph Angenendt -
Robert -
Robert