[Bug 446388] New: zypper package installation is delayed by nfs stat failure
https://bugzilla.novell.com/show_bug.cgi?id=446388 Summary: zypper package installation is delayed by nfs stat failure Product: openSUSE 11.1 Version: Factory Platform: x86 OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Network AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: mrmazda@ij.net QAContact: qa@suse.de Found By: --- Created an attachment (id=253226) --> (https://bugzilla.novell.com/attachment.cgi?id=253226) /var/log/zypp/history I'm filing here because factory mailing list anecdotes like this seem to get ignored. On none of my systems is any system software installed on anything but local filesystems. That means even though there may be active nfs mounts, there's no reason whatsoever I can think of for upgrade processes to stat them. Usually these systems have 4 exports from my local file server mounted, including during smart or zypper updating, which I normally do in runlevel 3. Today on one of my systems zypper dup was running along normally until I reached bug 446344 , after which the delay didn't cease even though the kernels were done installing. After about four hours elapsed to install the next relatively few packages I noticed that the downloading was proceeding normally, but the installations were all hanging. Eventually I got tired of waiting, expecting mirror updates to obsolete zypper's ref data before the end of the enormous time required to do the last 100 or so packages, so I did Ctrl-C. When zypper noticed, I got a cannot stat error on an NFS mount. My only guess what happened is that it has to do with zypper replacing portmap with rpcbind. After the zypper abort I did umount -t nfs -a, then redid zypper dup, and zypper proceeded to behave normally, quickly finishing the last 99 packages. Then I tried to mount the nfs shares again, and got bug 431722 errors instead of success, after which I did rcrpcbind, tried nfs mounting again, and succeeded. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 Andreas Jaeger <aj@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|bnc-team-screening@forge.provo.novell.com |jkupec@novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 User jkupec@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=446388#c1 Ján Kupec <jkupec@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|jkupec@novell.com |bnc-team-screening@forge.provo.novell.com --- Comment #1 from Ján Kupec <jkupec@novell.com> 2008-11-27 08:38:36 MST --- This is most probably not a zypper bug. I can assure you it does not touch any of your NFS mounts unless one of your repositories is located on them (or the directories like /tmp, /var/cache/zypp, /var/adm/, etc are on those mounts). Maybe some of the nfs mounts are in your $PATH or other variable that gets searched by various parts of the system? What exactly was the stat error you got? Please paste it here. Also attach zypper.log for whatever it might be worth. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 Ján Kupec <jkupec@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jkupec@novell.com Status|NEW |NEEDINFO Info Provider| |mrmazda@ij.net -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 User jkupec@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=446388#c2 Ján Kupec <jkupec@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |aj@novell.com Status|NEEDINFO |NEW Info Provider|mrmazda@ij.net | --- Comment #2 from Ján Kupec <jkupec@novell.com> 2008-11-27 08:43:45 MST --- Oh.. now i realized why you attached the history file :O) 2008-11-18 20:19:36|install|kernel-pae-extra|2.6.27.5-2.2|i586||openSUSE_Factory|35b3f3b04f93e87f3b0c71409000130c7111823e # 2008-11-18 20:31:06 kernel-default-extra-2.6.27.5-2.2.i586.rpm installed ok # Additional rpm output: # error: failed to stat /nfs/a-865/home: Input/output error ..and this happens for many more subsequent packages. I have no idea why, though.. but still it's not a zypper bug. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 User nfbrown@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=446388#c4 Neil Brown <nfbrown@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |nfbrown@novell.com AssignedTo|nfbrown@novell.com |mkoenig@novell.com --- Comment #4 from Neil Brown <nfbrown@novell.com> 2008-11-27 13:23:24 MST --- Clearly something is stating /nfs. Getting IO errors at this point probably isn't a big problem, but the hang clearly is. NFS should survive if portmap disappears on the client. Locking might have problems but everything else should work properly. Matthias: could you look at this. I don't know what happens when rpcbind gets installed, but if it is killing off any running portmap and starting an rpcbind, then that would be a problem as all the registrations would be lost. You need to either leave portmap running until reboot, or use rpm_dump to save the table, then pmap_set to restore it. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 User mkoenig@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=446388#c5 Matthias Koenig <mkoenig@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED --- Comment #5 from Matthias Koenig <mkoenig@novell.com> 2008-11-28 11:48:28 MST --- We cannot leave portmap running, because 1. this is done in the portmap package on uninstall and 2. rpcbind will not start correctly. But dumping the portmap tables and setting them after the rpcbind start would be an option. I also found, that currently rpcbind is not even started when portmap was running, this needs to be fixed. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=446388 User mkoenig@novell.com added comment https://bugzilla.novell.com/show_bug.cgi?id=446388#c6 Matthias Koenig <mkoenig@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED --- Comment #6 from Matthias Koenig <mkoenig@novell.com> 2008-12-02 07:32:05 MST --- Fixed in STABLE (should be in RC1) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com