[SOLVED] Another graph gaps thread

toddritt · Post by **toddritt** » Thu Dec 06, 2007 2:18 pm

I searched the forum (so don't flame me) but couldn't find a thread describing my exact problem, so I posted a new thread rather than hijack another one.

Anyway, Cacti was running fine for months monitoring my network, then one day the server hung (may have been a power outage, i don't remember now), and now every so many minutes there is a gap in all of my graphs (at the same time on all graphs). I upgraded to 0.8.7a to make sure my version didn't have a bug or something, but the same symptoms occur. I've attached the cacti log which clearly show a time out and a cron out of sync problem which I don't understand how to use. I'm using the CactiEZ cd with 0.8.7a cacti now (remember it worked fine at the previous version for months).

Hopefully my attachments are readable.

Thanks for any help that can be provided!

Todd

Post by **gandalf** » Sat Dec 08, 2007 8:24 am

Please try switching to cmd.php for some hours and report your findings
Reinhard

toddritt · Post by **toddritt** » Mon Dec 10, 2007 12:39 pm

Attached are the similar files while running cmd.php all morning. Thank you very much for your help.

Todd

Post by **gandalf** » Sat Dec 15, 2007 1:15 pm

There's a gap of 25 minutes in that log. Did you delete some lines? If not, please verify crontab execution from crond log or system log
Reinhard

toddritt · Post by **toddritt** » Mon Dec 17, 2007 10:01 am

I did not remove any lines (I just changed the username in the "authenticated" line to hide my true username).

There are no logs in /var/log/cron (where the cron is supposed to be logged based on my Webmin config...I use CactiEZ). Attached are screenshots of my scheduled cron jobs and the particular poller.php job which I figure may help.

Thanks for the assistance.

Todd

Post by **gandalf** » Wed Dec 19, 2007 3:03 pm

is this running on a vmware host?
Reinhard

toddritt · Post by **toddritt** » Thu Dec 20, 2007 12:39 pm

Yes it is on a VMware Server. Sorry I didn't say that earlier (but for what it's worth it was running fine for months).

Thanks!
Todd

Post by **gandalf** » Thu Dec 20, 2007 3:03 pm

Please search for similar threads. It's an issue with the timer running haywire.
Reinhard

Post by **TheWitness** » Thu Dec 20, 2007 10:38 pm

Gandolf is correct. VMWare has an issue on SMP boxes where the Guest OS' date will move ahead over time. Sometimes as much as 30 minutes to the hour. When either NTP or VMWares own clock synchronization takes effect, it essentially blows up the RRDtool update process as you can not update RRD's in the past.

The Way to get this is two fold:

1) Only use VMWare Apporved Kernels
2) Search for a 100Hz Kernel on the Internet

TheWitness

toddritt · Post by **toddritt** » Fri Dec 21, 2007 10:12 am

Thank you both for your help! I searched the thread for the kernel issue, found a 100HZ centos kernel, figured out how to install it, etc., and so far I've seen an immediate remedy with my graphs working normally now.

Much appreciated,
Todd

[SOLVED] Another graph gaps thread

[SOLVED] Another graph gaps thread

Who is online