On 01/28/2010 10:01 AM, Ton Voon wrote: > > On 17 Dec 2009, at 08:12, Brandino Andreas wrote: > >> I have much less hosts and services (for the moment). >> After deleting "retention.dat" I still face the same delay... >> > > When did this slow down occur? I can see how Jonathan's patches speed > up the system but these look like long-term bottlenecks, not a 3.0 -> > 3.2 migration. I'm interested to know when the slowdown occurred and > if it is the side effect of something else. > > There was a patch applied sometime in 3.1 which improved the circular > parents lookup, but I don't know much else (Andreas applied it but > didn't update the Changelog). >
I don't do changelogs :p The info is in the commit history though. I think you're referring to the following patch, sent in by Jean Gabès: commit a6e06d8de24ffcb4a8c341a60098042dc3284756 Author: Andreas Ericsson <[email protected]> Date: Sun May 17 12:54:28 2009 +0000 Revamp hosts' circular parent/child detection With this patch, all hosts are checked at most twice when determining circular host/parent relationship chains. Previously, all hosts and all their parents were always checked once for each host, which caused for the non-linear scaling function represtented as O((n*n)+depth_of_n). For some timing comparisons: A configuration with 151109 nodes is used. The 150300 interesting nodes are in chains 300 levels deep, with 501 hosts in each level. Patched: 6.28user 0.19system 0:06.58elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+65723minor)pagefaults 0swaps Unpatched: (Ctrl-C'd out as it wasn't done after nearly an hour) 3221.77user 0.31system 53:56.51elapsed 99%CPU (0text+0data 0max)k 0inputs+0outputs (0major+67480minor)pagefaults 0swaps Without even completing the timing comparison, the patch provides a speedup of more than 51300%. On a smaller and more real-worldy configuration of roughly 15000 hosts, we get these timings: Patched: 0.71user 0.02system 0:00.75elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k 0inputs+0outputs (0major+8219minor)pagefaults 0swaps Unpatched: 213.77user 0.02system 3:34.34elapsed 99%CPU (0text+0data 0max)k 0inputs+0outputs (0major+8280minor)pagefaults 0swaps A huge improvement indeed, and one that's necessary for Nagios to be usable in truly huge networks. Signed-off-by: Andreas Ericsson <[email protected]> -- Andreas Ericsson [email protected] OP5 AB www.op5.se Tel: +46 8-230225 Fax: +46 8-230231 Considering the successes of the wars on alcohol, poverty, drugs and terror, I think we should give some serious thought to declaring war on peace. ------------------------------------------------------------------------------ The Planet: dedicated and managed hosting, cloud storage, colocation Stay online with enterprise data centers and the best network in the business Choose flexible plans and management services without long-term contracts Personal 24x7 support from experience hosting pros just a phone call away. http://p.sf.net/sfu/theplanet-com _______________________________________________ Nagios-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nagios-users ::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. ::: Messages without supporting info will risk being sent to /dev/null
