[3945] in Release_7.7_team
Proposed solution for RPM hangs
daemon@ATHENA.MIT.EDU (Greg Hudson)
Mon Jul 21 04:48:08 2003
Date: Mon, 21 Jul 2003 04:48:04 -0400
Message-Id: <200307210848.h6L8m4Wf003691@error-messages.mit.edu>
From: Greg Hudson <ghudson@MIT.EDU>
To: release-team@MIT.EDU
We've had two field reports of updates hanging, one from Garry and one
from vchudnov. I think we need to address this problem sooner or
suffer a lot later. (We're already doomed to suffer a little, but
hopefully not very much.)
My proposed solution is to grab the rpms from
ftp.rpm.org:pub/rpm/test-4.2, put them in a special subdir of
redhat-9-updates, and add those to the release. They appear to be at
a stable point right now and are supposed to fix all known hanging
problems. An errata is supposedly forthcoming from Red Hat, but I've
been holding my breath for several months now and it hasn't helped.
Background: the problem is a locking issue in rpmlib. The manual
remedy is to kill -9 all rpm/rpmupdate processes and run "rm
/var/lib/rpm/__db.*", but we can't expect our users (or our public
machines) to know that.
http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=73097 is the
"master bug" describing this issue, which is actually a collection of
smaller issues (any circumstance which can cause rpm to terminate and
leave behind a stale lock). You can see how Jeff Johnson promises an
errata on 2003-03-08 but it never arrives, and you can see him
recommend the test-4.2 RPMs there.
http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=88164 also
contains a comment from Jeff Johnson recommending the test-4.2 RPMs on
2003-04-14.
