[dokuwiki-teams] Re: [infrastructure] server load
- From: "Dennis Plöger" <dennis@xxxxxxxxxxxxxx>
- To: dokuwiki-teams@xxxxxxxxxxxxx
- Date: Thu, 03 Mar 2011 18:09:50 +0100 (CET)
Hi!
Looking at the Zabbix graphics (see attached image), I doubt, that the backup's
doing peaks, because the peaks are too irregular for the backup.
But about the wiki: I don't know, where our wiki could produce high cpu loads.
In my experience, high cpu loads are high i/o waits, which I doubt for or small
wiki and the (I hope) fast raid in our server. The other thing is high
calculations or image processing, which don't count for the wiki, too, because
there ain't any like this.
What does the access or error log say?
Dennis
P.S.: I finetuned the zabbix error mails a bit as it was sending doubled mails.
If we finally see, that the high loads are okay, I can update the trigger to
reflect that and be quiet in those times.
Andreas Gohr <andi@xxxxxxxxxxxxxx> wrote ..
> Hi,
>
> I noticed that we get somewhat frequent alerts for high load and process
count.
> http://dib.dokuwiki.org/munin/localdomain/localhost.localdomain-cpubyuser.html
> suggests some really bad peaks. The root peaks are probably backup
> jobs? The wiki peaks on the other hand might be a serious problem. Any
> ideas how we could find out what's causing them?
>
> Additionally, could we change the alerts to not send mails immediately
> but only when a problem persists for, let's say a minute or so?
>
> Andi
>
> --
> splitbrain.org
Other related posts:
