Gaps in Graphs since upgrading to v0.8.6b on FC2

Post by **TheWitness** » Thu Dec 02, 2004 5:23 pm

Here is the picture of "Settings" showing debug.

_JPL_ · Post by **_JPL_** » Fri Dec 03, 2004 8:14 am

THE SOLUTION TO MY PROBLEM:

Installing the FC2 rpm created a second cron entry and a second user for running cacti called 'cacti'. When the 2 cron entries overlapped their runtime it produced gaps in the graphs. I removed the user and cron entry created by the FC2 rpm and now all is working right again.

Post by **rony** » Fri Dec 03, 2004 9:15 am

I was just speculating that last night with Ian......

Glad that was it.

castor · Post by **castor** » Fri Dec 03, 2004 1:05 pm

seems like it was the same thing here :p
greets !!!

anyway i did'nt found where the cron task by cacti user is configured, but i disabled the one i put myself in /etc/crontab and it seems to wrok nice for the moment.

i have a question about the cacti user, why is it configured with a /bin/bash shell rather than /sbin/nologin ?

_JPL_ · Post by **_JPL_** » Fri Dec 03, 2004 7:10 pm

I have mine as /bin/false I think.

It doesn't need to be a shell so you should change that. For me the cron entry was put into /etc/cron.d and it was called cacti as well as the corresponding user. That should be the same for everyone if they installed the FC2 rpm.

Post by **rony** » Fri Dec 03, 2004 8:54 pm

You could also just do a

Code: Select all

crontab -e -u cactiuser

as root and edit the users crontab to run it.

dorje · Post by **dorje** » Wed Dec 08, 2004 9:33 pm

rony / _JPL_,

I believe I have a related problem.
Being new to Linux could you let me know how to check if I have two cacti users (also run FC2 installed from rpm) ?

I have one 'cacti' entry in my /etc/cron.d/ folder.
_JPL_ is this the cron entry I should remove ?

Should I do that by using rony's code ?

Code: Select all

crontab -e -u cacti

Further to this _JPL_ you mention actually removing the user as well. What is the best way to do this ?

Cheers from the South Pacific.
DM

_JPL_ · Post by **_JPL_** » Mon Dec 13, 2004 10:46 am

Check in /etc/passwd for more than one cactiuser. If you created one manually and called it something other than 'cacti' then you should have 2.

For me, I removed the /etc/cron.d/cacti file because I had it already defined in 'cactiuser' crontab. Simply deleteing it will do.

dorje · Post by **dorje** » Mon Dec 13, 2004 2:51 pm

_JPL_ many thanks for that - unfortunately I've only got one Cacti user.
Cheers for the post - I'll keep looking.

DM

Arnar · Post by **Arnar** » Fri Dec 17, 2004 6:08 am

I seem to be having the same problem as _JPL_ and Castor. Like Castor I have entries in my log stating that the poller exceeded the 293 second limit, but when I run it by hand it only takes about 7 seconds. I'm not using cactid.

The gaps in my graphs are very frequent, about 2-4 each hour, honestly, they look like barcodes

)

I'm running 0.8.6b with rrdtool 1.0.37 and NET-SNMP 5.0.1 on Slackware 8.1.

This seems to happen to all my graphs but one, which is local (linux) memory usage. The other graphs are mostly generated from snmp data, except one which is from a local script on the cacti host. That graph also has gaps in it.

The other hosts that I am polling are two Windows 2000 server machines, two Netware servers and two Linux machines. The linux machines are running NET-SNMP 5.2

I'm attaching some of my log file, after I turned on debugging, this happens for the run at 10:50 on 12/17/2004 (see note).

Arnar

note: a thought: maybe ISO Y-M-D date format would be better for log files rather than the US-specific M/D/Y.

Arnar · Post by **Arnar** » Fri Dec 17, 2004 6:09 am

I forgot to tell you that I already checked my crontabs and I have only one user (nobody) running poller.php each 5 minutes.

thanks,
Arnar

_JPL_ · Post by **_JPL_** » Tue Jan 04, 2005 3:08 pm

Anything fishy in /var/log/cron?

Morgan · Post by **Morgan** » Wed Jan 05, 2005 3:54 am

i'm having the same problem guys.

cacti -> cacti-0.8.6c/

not an rpm package, ( i use tarballs)

PHP 4.3.10
rrdtool-1.0.49
NET-SNMP version: 5.1.1

the gaps i'm seeing are ONLY on interface graphs for bits per sec + total bw, and is happeneing on cisco routers, located far from my poller ( asia or south american primarily or the ones gaping)

bluecoat proxy is not gaping, other weird snmp queries done by hand arent gaping, scripts arent gaping.

Post by **rony** » Wed Jan 05, 2005 9:06 am

Morgan,

Try increasing the snmp timeout for that host or the global one. If I remember correctly it is less than previous version or it accually being enforced now.

But Larry (TheWitness) can answer more accuratly than I can.

Morgan · Post by **Morgan** » Wed Jan 05, 2005 12:07 pm

Larry any thoughts?

ps: just changed the snmp timeout from 500 to 1000

will see if it makes any difference overnight but i doubt that is the cause.

M

Cacti

Gaps in Graphs since upgrading to v0.8.6b on FC2

Newbie

Many thanks

Gaps in all graphs

I forgot..

Who is online