Cacti GAPS in Disk Usage when Monitoring Win32 Hosts

mrh · Post by **mrh** » Tue Mar 08, 2005 10:55 am

Hi,
I've a veri weird issue when monitoring some win32 hosts with cacti. The disk usage is not properly shown, but only one graph per host, so if I delete this graph and back to create, it will show correct values, but the next graph in the disk usage template gets corrupt and begins to show incorrect values.

I've tried all: emptying the poller cache, deleting the entire host and creating it back... changing the rrdtool version... reinstalling the SNMP service in the machines affected...

Please, I will thank a lot any info provided of how to solve this.

MRH

DevilSun · Post by **DevilSun** » Tue Mar 08, 2005 11:01 am

How is it you're querying that data, are you using the WMI script posted before or actual SNMP values? I'm using a perl based WMI script and I don't have any of those on any of our servers...

Post by **rony** » Tue Mar 08, 2005 11:27 am

Increase you SNMP timeout for those hosts.

pbulteel · Post by **pbulteel** » Tue Mar 08, 2005 11:56 am

I had this behaviour too and couldn't figure out what was wrong. I didn't change the timeouts but after deleting my 500 meg cacti log file everything seemed to perform better.

So check how big your log file is.

-P

mrh · Post by **mrh** » Tue Mar 08, 2005 12:07 pm

Thanks for your replies, I've put the SNMP timeout to 5000 and the problem still continues. Regarding my logfile, I've cleared it too many times, also the cache but nothing good happens.

Devilsun, my queries are all SNMP. I don't know by now how to do WMI queries

MRH

pbulteel · Post by **pbulteel** » Tue Mar 08, 2005 12:26 pm

By any chance have you switched between using cmd.php and cactid for polling?

-p

mrh · Post by **mrh** » Tue Mar 08, 2005 1:08 pm

not yet pbulteel, I tried to do that but I found cactid must be compiled before and I'm not sure how to do that... Do you know if there is a comprehensive guide for doing this in an easy way?

mrh

mrh · Post by **mrh** » Tue Mar 08, 2005 1:26 pm

...well, I suppose if I've been able to install cacti I'll be able to install cactid too... I'll try and tell you if my problems gone! It's strange, because my total poll completes in about 5 secs, just monitoring 5 servers in a local 100mbps network connection.

mrh

mrh · Post by **mrh** » Mon Mar 14, 2005 6:39 am

nothing, finally it's NOT working, very weird because it's happening only in certain hosts. I'm currently monitoring about 10 2000, XP and 2003 machines and this is completely random. I've reinstalled the SNMP service, chkdsk to all partitions... Even installing cacti from zero in another PC

Only to say cacti developers to look into this and fix it if possible in a future release

mrh · Post by **mrh** » Mon Mar 14, 2005 6:41 am

(attached file regarding last post, very weirs because the periodicity of the spikes!!)

Alives · Post by **Alives** » Thu Jul 14, 2005 4:49 pm

Im having the same problems here and only on one of my hosts. I am checking from linux so i cannot use WMI. I am using cmd.php and I think i will install cactid now just to test...besides my poller is nearing 300 sec.

I just increased the snmp timeout so we'll see how that works out.

Thanks!!

Post by **TheWitness** » Thu Jul 14, 2005 5:22 pm

Per my other post. Pre 0.8.6e HostMib queries do not respect the timeout value. In addition, due to some brain fart on the part of Microsoft, when polling the HostDisk stuff, the A: drive is polled causing long responses.

Therefore, you need to verify that you are using the latest (0.8.6e/f) scripts and resources files. Also, increase your snmp timeout to between 5-8 seconds.

Alternatively, remove your polled devices A: drive

. Or better yet Call Bill Gates personaly and complain about his lack of adopting industry standards

TheWitness

Alives · Post by **Alives** » Fri Jul 15, 2005 10:31 am

that fixed it... i made it 10 secs for the hell of it...

THANKS A TON!!!!

ioiioi · Post by **ioiioi** » Thu Sep 15, 2005 7:38 am

unforturnaly, even though increasing the snmp_timeout to 10 secs. My unconsistance situation remain the same, maybe I should try to use w32 wmi for monitoring disk capacity.

mourn · Post by **mourn** » Thu Sep 15, 2005 11:04 am

I'm having the same issue. Changing timeout values doesn't seem to work and it only affects the C: drive on my testing host, the E: drive graph is fine. Very odd.

Cacti GAPS in Disk Usage when Monitoring Win32 Hosts

Cacti GAPS in Disk Usage when Monitoring Win32 Hosts

Who is online