Red Hat Bugzilla – Bug 189318
yum reliably triggers oom-killer
Last modified: 2014-01-21 17:53:53 EST
Description of problem: Yum suddenly started triggering the oom-killer today. Version-Release number of selected component (if applicable): I tried sever versions on yum from latest repo-yum as/of few days ago to yum-2.4.2-69.rhfc4.at.x86_64.rpm & yum-2.6.0-1.src.rpm from FC5 sources I tried a couple of kernels (stock FC4, x86_64) 2.6.16-1.2069_FC4 2.6.15-1.1833_FC4 How reproducible: Every time since just today. Steps to Reproduce: 1. yum info yum (actually seems independent of args) Actual results: 2. after 5 seconds-or so, starts using swap 3. after few-more secs starts oom-kill thrashing 4. system DOES respond to SysReq S/U/B recipe Expected results: <heh> shouldn't gobble all the world's memory Additional info: from cat cpuinfo.. AuthenticAMD cpu family : 15 model : 47 model name : AMD Athlon(tm) 64 Processor 3500+ stepping : 0 cpu MHz : 2210.202 cache size : 512 KB Normal memory stats: free -m total used free shared buffers cached Mem: 1004 580 423 0 25 282 -/+ buffers/cache: 272 732 Swap: 972 0 972 Most recent yum updates Apr 13 23:13:53 Updated: plone.x86_64 2.1.2-2.fc4 Apr 13 23:14:43 Updated: abiword.x86_64 1:2.4.4-2.fc4 Apr 14 19:42:47 Updated: gnochm.noarch 0.9.7-4.fc4 Most recent manual installs (reverse-chron): python-sqlite2-2.1.3-4 sqlite-3.3.3-1.2 lighttpd-1.4.11-1.x86_64.rpm python-urlgrabber-2.9.8-2.noarch.rpm linkchecker-3.4-1.x86_64.rpm
why'd you install sqlite 3.3.3? b/c that's what's causing the memory overrun.
Whoa! How did you pinpoint that? That seemed to be it. I did a rpm --oldpackage back to sqlite-3.1.2-3 (from FC4) and yum works now. To answer your question: I had some thought of upgrading python-sqlite2, and seemed to <accidentally> grab the 2.2.1.3 version from FC5. Then when it griped about needing sqlite3.3, I thought I would be safe installing the FC5 sqlite3.3 if I compiled it in my environment -- guess that may have been an incorrect assumption <grin>. So ... problem is resolved.. ..but I still want to know how you knew? My sincere respects!! And warm thanks for the prompt feedback, too. ..jim
It's been reported before. Paul Nasrat discovered it. I just read it. :)