Red Hat Bugzilla – Bug 612292
Performance loss querying object database from command line
Last modified: 2016-04-26 10:19:49 EDT
Description of problem: In order to report errors with /dev/shm being full, writes zeroes to all of the pages. This was done to resolve bug 607738. This introduces a performance regression for utilities which query a couple of things from the object database (ex: rgmanager's cmdline utility 'clulog'). Version-Release number of selected component (if applicable): 1.2.3-9 How reproducible: 100% Steps to Reproduce: 1. time clulog -s7 hi Actual results: 1+ second per run Expected results: faster ;) Additional info: clulog is not the only piece of software affected; it just happens to be the one I noticed. A couple of cursory tests with Steve showed that increasing the buffer size to 4096 had a significant improvement, while values larger than that had little improvement.
Created attachment 430185 [details] flatiron revision 2987 to fix problem
[root@crackle ~]# time clulog -s7 hi hi real 0m0.211s user 0m0.007s sys 0m0.070s [root@crackle ~]# rpm -q corosync corosynclib corosynclib-devel corosync-1.2.3-12.el6.x86_64 corosynclib-1.2.3-12.el6.x86_64 corosynclib-devel-1.2.3-12.el6.x86_64 Works for me.
Red Hat Enterprise Linux 6.0 is now available and should resolve the problem described in this bug report. This report is therefore being closed with a resolution of CURRENTRELEASE. You may reopen this bug report if the solution does not work for you.