On 10 Dec 2009, at 14:32, John Hodrien wrote: > On Thu, 10 Dec 2009, Vanush Misha Paturyan wrote: > >> On Tue, 8 Dec 2009 11:04:47 +0000 (GMT) John Hodrien >> <[email protected]> wrote: >> >>> Watching the disk array this morning, it appeared that you'd get a >>> period of normal blinking activity lights, then it'd all go silent >>> for maybe 5 seconds, then solid on for a short while, then back to >>> blinking. This looked odd to me. >> >> I find "sysstat" http://pagesperso-orange.fr/sebastien.godard/ >> package >> with it's "sar" commands and crontab jobs very handy to collect >> information and analyze it later. I run Debian on servers so for me >> simple "apt-get install sysstat" and afterwards modifying >> "/etc/default/sysstat" to allow for data collection is sufficient to >> get files in /var/log/sysstat. Then wait for system to misbehave and >> start analyzing data with "sar" command. Whatch out for memory >> consumption and where does CPU spend most of the time? > > Yep, I'm using sar. Memory consumption isn't really an issue (it's > not > swapping in the slightest, in fact about 6.5 of the 8Gbytes it has is > accounted for by 'cached'. While normally idle it reports 0.04% > average > iowait, but this rises to ~50% under this particular nfs load. > > jh > >
What about block device activity (-d option)? Any noticeable difference in average size of the requests or I/O wait time between normal and NFS writes? Also, stalls only happened during NFS writes maybe it's not a storage issue but a network card/driver issue (locking up kernel on a high network traffic?) This is more of a brainstorm session suggestions than actual advice how to fix problems though. Regards, Misha --- Vanush "Misha" Paturyan Senior Technical Officer Computer Science Department NUI Maynooth _______________________________________________ Linux-PowerEdge mailing list [email protected] https://lists.us.dell.com/mailman/listinfo/linux-poweredge Please read the FAQ at http://lists.us.dell.com/faq
