help snmp error

vectorz · Post by **vectorz** » Wed Jan 24, 2007 5:42 pm

I'm trying to use cacti to graph some custom OID's from Dynamo's mib. Dynamo (appserver) has its own snmp agent built in on port 8270.

So i configured a device to use snmp v1 -c public port 8270 and 'generic snmp enabled host', however, I keep getting 'snmp error' when it tries to poll, and the logfile keeps spitting out 'result from SNMP not valid'.

From the cacti server, if i run the actual command to do snmpget -v1 -c public 192.168.1.30:8270 1.3.6.1.4.1.2726.1, I get a valid return so I know all the settings are correct.

I read up on the snmp error stickies in the HOWTO section but I'm not making much progress.

Can anyone tell me what I'm doing wrong?

BSOD2600 · Post by **BSOD2600** » Wed Jan 24, 2007 9:43 pm

1) does that device respond to sysuptime / system ?
2) does cacti consider the device up?
3) change the logging level to medium and watch what cacti returns for that device. paste the results.

vectorz · Post by **vectorz** » Thu Jan 25, 2007 1:21 pm

Thanks for the response...

The dynamo snmp agent doesn't respond to sysuptime nor system, because its mib doesn't have those oids.

Cacti does not consider the device up, I believe because the poller is set to snmp for that.

I changed the logging level all the way up to debug, and now realize the 'result from SNMP not valid' errors do not pertain to this particular dynamo monitor. They're related to the other working monitors. Looking through the log, I dont' see any reference to the new dynamo device at all, yet I'm sure the device is not set to 'disable host'.

Post by **gandalf** » Thu Jan 25, 2007 2:04 pm

Please try latest cacti 0.8.6j. There was a change on Downed Host Detection. It will now do an snmpgetnext against OID ".1".
Reinhard

vectorz · Post by **vectorz** » Thu Jan 25, 2007 2:10 pm

Roger that, thanks for the heads up!

vectorz · Post by **vectorz** » Thu Jan 25, 2007 4:13 pm

I upgraded to 0.8.6j (using the upgrade documentation), and now I see the following in the cacti.log, but still getting "SNMP Error".

** I verified this server can resolve server hostname no problems.

01/25/2007 01:10:35 PM - CMDPHP: Poller[0] Host[86] SNMP: Host did not respond to SNMP
01/25/2007 01:10:35 PM - CMDPHP: Poller[0] DEBUG: SQL Exec: "update host set status = '1', status_event_count = '1', status_fa
il_date = '2007-01-19 03:15:31', status_rec_date = '2007-01-25 01:10:05', status_last_error = 'Host did not respond to SNMP',
min_time = '9.99999', max_time = '1187.01000', cur_time = '152.64000', avg_time = '980.92231', total_polls = '44913', failed
_polls = '2262', availability = '94.963596286153' where hostname = 'server1.mydomain.com'"

Post by **gandalf** » Thu Jan 25, 2007 4:42 pm

Are you using php-snmp? Click the cactus on the main page and scroll to the bottom to find it. What are your Settings for Downed Host Detection
Reinhard

vectorz · Post by **vectorz** » Thu Jan 25, 2007 5:42 pm

Hmm... that could be the prob. I thought I was supposed to use net-snmpd but oddly it's been workign w/ all other snmpd linux type of monitoring and cisco stuff.

PHP SNMP Support: no

Poller Type: cmd.php

BSOD2600 · Post by **BSOD2600** » Thu Jan 25, 2007 8:16 pm

Enable snmp support for php by adding the php_snmp module in the php config file.

tl1997 · Post by **tl1997** » Sat Feb 03, 2007 1:17 am

gandalf wrote:Please try latest cacti 0.8.6j. There was a change on Downed Host Detection. It will now do an snmpgetnext against OID ".1".
Reinhard

This new host detection is not working for me. I upgraded to 0.8.6j from 0.8.6i and now Juniper routers and PowerWare UPSes don't work anymore. My config is unchanged. Readding the host doesn't help. It seems obvious why the PowerWare check doesn't work, but the Juniper has me a bit confused, since it responds to .1, but the debug in cacti.log shows "host did not respond to SNMP" in both cases and skips it.

Juniper T320:

% snmpgetnext -c public -v 1 x.x.x.x .1
iso.2.840.10006.300.43.1.3.0 = Timeticks: (0) 0:00:00.00

PowerWare UPS:

% snmpgetnext -c public -v 1 x.x.x.x .1
Timeout: No Response from x.x.x.x.

Any ideas?
Anyone else have this problem?
Is this a bug or a feature?

Thanks!

vectorz · Post by **vectorz** » Sat Feb 03, 2007 1:56 am

I gave up.

Post by **TheWitness** » Sat Feb 03, 2007 9:29 am

1) What version of PHP?
2) Does the UPS respond to get's for either sysDescr or sysUptime?
3) Have you applied the patches at the web site?
4) Does the APC respond to anything?

If the answer to 1) is 4.x, please apply the patches from the website

If the answer to 2) is yes, change your ping failure count to more than 1 say 3.

If the answer to 3) is no, please do so.

If the answer to 4) is no, fix that after you apply the patches.

TheWitness

Post by **TheWitness** » Sat Feb 03, 2007 9:36 am

For the appserver, try the following:

snmpgetnext -c <your_community> -v <your_version> <your_hostname>:<your_port> .1

If it responds correctly, you need to follow the instructions in the prior post and you should be Ok.

TheWitness

tl1997 · Post by **tl1997** » Sat Feb 03, 2007 2:47 pm

TheWitness wrote:1) What version of PHP? TheWitness

5.2.0

TheWitness wrote: 2) Does the UPS respond to get's for either sysDescr or sysUptime?
TheWitness

Yes

TheWitness wrote: 3) Have you applied the patches at the web site?
TheWitness

No...I assume you are referring to the patches that restore the phpv4 compatible checking? I was hesitant to do so, since all 70 other some odd hosts (that aren't PowerWare) work fine.

TheWitness wrote: 4) Does the APC respond to anything?
TheWitness

PowerWare actually

Yes, it does. I found a post in another thread were some guy changed line 284 (I think) of lib/ping.php from .1 to .1.3 which fixed my problem and now works for all hosts.

TheWitness wrote: If the answer to 1) is 4.x, please apply the patches from the website

If the answer to 2) is yes, change your ping failure count to more than 1 say 3.
TheWitness

I didn't end up doing this since the change to .1.3 worked. Should I change it back and do this? Does that just cause it skip ahead down the OID tree? Or just retry the same thing?

Was this to fix the Juniper or the Powerware or both?

tl1997 · Post by **tl1997** » Sat Feb 03, 2007 2:51 pm

TheWitness wrote:For the appserver, try the following:

snmpgetnext -c <your_community> -v <your_version> <your_hostname>:<your_port> .1

If it responds correctly, you need to follow the instructions in the prior post and you should be Ok.

TheWitness

See my previous post...I did the same thing except omitted the port number since its standard. The Juniper responds but the PowerWare does not. Based on the post about the same problem the guy with the MGE UPS was having, I changed to .1.3 in lib/ping.php which works. Still not sure if I should be making this change. I try to not to apply changes that'll I'll lose in the next upgrade.

Thanks for your help.

help snmp error

help snmp error

Who is online