Some drive usage graphs are incomplete, help!!

[jcullen]

Thanks to all in advance!!!

some of my grpahs are not complete for hard drive usage. I have tried to delete the rrd files associated with the graphs with no help. Only about 5-10% of the graphs are not working and I am having problems isolating the culprit. I am getting either:

do data at all
or one of the two: Capacity or Used

I am a newbie to troubleshooting cacti issues, so thanks for the help and sorry if I could not find the post that might have answered this.

[melchandra]

Is there anything specific about the graphs that don't work? Are all the hosts running linux? or windows? or something else. A good tool to verify snmp functionality is GetIF. You can use it to tell if it's a snmp issue with your hosts, or somethign else. Cacti should automatically re-create rrd files, but that doesn't necessarily mean it will. Something else might have gotten screwed up. You might try deleting the offending hosts and re-adding them if you're not worried about loosing past data or anything.

[jcullen]

What we are doing is just simple hard drive capacity-usage graph. All of the problems are windows specific. Most servers have 3 drives at min and in a C, D, E set up. E is usually the bigger drive ranging from 60GB to 100GB. The E drive is the drive usually having the issues. Although some virtual memory graphs are messed up in the same way.


***Also, since we first starting using cacti some of out interfaces have changed from not being used to up and named. Now they are not pulling the correct name. What is the easy way to do this without losing traffic data. If there is no easy way to do that I am willing to kiss those bits good bye. Just that some of those are for clients.

THANKS for the heads up on GetIF going to look into it
<-- Linux Newbie

[jcullen]

Well now that I have got GetIF installed and using it is not helping me too much in finding the culprit (probably me not knowing how to work with the MIBs well). I want to try and findout where the wrong data is coming from and why. On my graphs I am getting:

nan

when there should be xx.xGB,

where should I look for the problem?


Anyone one got any ideas on the Title of the graphs:

I use "|host_description| - Used Space - |query_hrStorageDescr|" for the Title of the graph

since we fist implemented some interfaces have had names applied to them and '|host_description| - Used Space - |query_hrStorageDescr|" still is showing up.

I would rather not lose the data but if I do it is not as big as an issue as just getting them fixed. Is there a way without deleting the host and readding??????

THANKS so MUCH ALL???

[melchandra]

Check the output of your cmd.php. (If you log it, or run it manually). Is it getting a value or reporting any kind of error? Do you have a large amount of graphs? Cause if cmd.php runs longer than 5 minutes in a single run, it can cause problems.

[jcullen]

These are some of the Standard error outputs of "php cmd.php"


Error in packet
Reason: (noSuchName) There is no such variable name in this MIB.
Failed object: interfaces.ifTable.ifEntry.ifOutOctets.59


ERROR: illegal attempt to update using time 1092277044 when last update time is 1092277044 (minimum one second step)


ERROR: could not lock RRD
ERROR: expected 1 data source readings (got 0) from Such:...

...Any ideas on where to go from here....thanks so much

[melchandra]

Did you try this more than once? It almost looks like you have multiple cmd.php processes running. I say that because of the "could not lock rrd" and the fact that it thought you were trying to update twice in the same second. Check a "ps -ef | grep cmd.php" and see if you have zombie cmd.php processes or anything. If not, then you need to figure out what has locked the rrd files (or did a lock just not get removed and you need to do it manually)

[jcullen]

ran 'ps -ef | grep cmd.php'

root 5383 32146 0 22:46 pts/1 00:00:00 grep cmd.php

the other two errors I get are more frequent, the above error only seems to happen once. Just to double check...this is the command I run
php cmd.php 2> log.cmd.php

what is the easiest way to find out the MIB or OID so I can use GetIF to see the actual value the snmp walk gets.

Anything else I can do to help.

Thanks Dave,

[melchandra]

In GetIF, on the start page, enter your hostname, and community string. Click start to test it, you should see all the sysinfo stuff pop up. If that works, go to the MBrowser tab. Here you should be able to walk everything. I would just click start and let it grab everything. And then you can browse the results on the bottom. Otherwise, just click the pluses to navigate your way down the tree. iso.org.dod.internet.mgmt.mib-2.interfaces is a good place to start.

Error in packet
Reason: (noSuchName) There is no such variable name in this MIB.
Failed object: interfaces.ifTable.ifEntry.ifOutOctets.59

The problem might me that your device got rebooted or something, and re-indexed the interfaces.

[Guest]

Thanks for your help, I am learning alot and still going!!! Now I am having problems trying to find the actual MIBs cacti is using for the (windows) hard drive usage. Is there a MIB file I need to load with GetIF? Also, most of my graphs I am having problems with is the E drive on 90% of one clients network. All other drives on these servers are responding fine. Most of them I am getting "nan" for the data portion. Some I am getting capacity just not usage. Any ideas????

THANKS SO MUCH. Glad to see someone will help.
Try to help one stranger a day!!!!

[melchandra]

You could try the host MIB.. I think that's what cacti uses by default. It's root oid is .1.3.6.1.2.1.25. More specifically, the storage tables located at .1.3.6.1.2.1.25.2.3.1.

As far as some working and some not... that's tricky. Try to identify what's differant about the one's that are working and the one's that aren't. Check the output of cmd.php for errors. Verify by looking at the timestamps that the rrd files really are getting updated.

[jcullen]

Sorry it took awhile to Post:

fix for incomplete hard drive graphs.

if you delete the data source for the particular graph on that host and just recreate it, it will start pulling the create data.

Not sure if this was a bug that was fixed in the upgrade or not.

Thanks to Dave who helped on this matter.