Device Status "Unknown"...why???

DevilSun · Post by **DevilSun** » Tue Sep 28, 2004 11:21 am

This is a device I had previously setup in 0.8.5 and had it disabled. I deleted it, and a few days later I re-added it now that I have 0.8.6. The device exists, is contactable via both pinging and SNMP yet shows up as "Unknown" status when I list devices and am not getting any graphs for it.

I have only the SNMP-Interface Stats Data Query setup for it, and the status of that dq is "Success [209 Items, 26 Rows]". I do a verbose query, and everything is correct. I added graphs to the device all fine and dandy but they're not getting created because the device is in an unknown state. I've disabled and re-enabled the device. Done everything I can think of, and it's still unknown after the device has been in Cacti overnight.

Poller settings are "Ping" because I have a few devices that don't respond to strict SNMP info (Win2003 hardened domain controllers), and the ping type was UDP. I just switched it to ICMP, but I don't see why this would be a problem.

What else could be causing this device to be "Unknown" status to Cacti?

Win2003 w/ IIS6
Cacti 0.8.6
MySQL 4.0.20a
PHP 4.3.7.7
Perl 5.8.4.810

DevilSun · Post by **DevilSun** » Tue Sep 28, 2004 11:26 am

Oh yeah, using the cmd.php poller too not cactid. Not having any problems with any other devices.

DevilSun · Post by **DevilSun** » Wed Sep 29, 2004 9:53 am

I'm sorry, is this too complex of a problem to get any input or do I happen to get the weird problems that no one has ever seen before?

DevilSun · Post by **DevilSun** » Thu Sep 30, 2004 5:40 pm

TTT once again...switch has been in Cacti and enabled for 3 days now, still showing up as "Unknown" even though the device is pingable and return SNMP data...

addlema · Post by **addlema** » Fri Oct 01, 2004 7:35 am

Actually i have this same sort of problem.. I have 3 devices that were listed as UP in Cacti, they then went down, and now have been up for 5 days but Cacti still lists them as down...

Also i have had a device in Cacti Recovering for about 3 days now...

Also 2 devices I added on Monday have not left here unknown status yet...

Cacti .0.8.6 on Gentoo

DevilSun · Post by **DevilSun** » Tue Oct 05, 2004 2:06 pm

...

Post by **TheWitness** » Tue Oct 05, 2004 5:24 pm

DevilSun,

Just read this one for the first time. I believe that there may be an issue with PHP that is causing this problem. Run cmd.php from the command line in LOW and see if you are getting PHP warning messages.

This is likely causing the problem. Post your results. Also, verify that the sockets extension is built into PHP if you are running Unix or the php_sockets.dll extension included in the php.ini file. This is required if you are using either UDP or ICMP.

Thanks,

TheWitness

DevilSun · Post by **DevilSun** » Wed Oct 06, 2004 1:45 pm

I have 14 hosts, however when I ran it under Medium so I could see results, it appears 1, if not 2 hosts are never showing up as being polled, i.e. doesn't seem to be gathering data for them which is fairly odd.

Zero warnings...PHP sockets is good to go as well.

Code: Select all

Content-type: text/html
X-Powered-By: PHP/4.3.7

#!/usr/bin/php -q
10/06/2004 11:43:24 AM - CMDPHP: Poller[0] Time: 55.2745 s, Theads: N/A, Hosts: 14

DevilSun · Post by **DevilSun** » Wed Oct 06, 2004 3:01 pm

Here's something even more weird, here's what's in my log file for poller.php

Code: Select all

10/06/2004 12:55:36 PM - SYSTEM STATS: Time: 36.3913 s, Method: cmd.php, Processes: 4, Threads: N/A, Hosts: 15, Hosts/Process: 4

cmd.php reports 14 hosts, poller.php says 15?

Post by **TheWitness** » Wed Oct 06, 2004 7:39 pm

DevilSun,

We have a temporary workaround in cmd.php to address non-host based data sources. This feature is depricated but was supported in the earliest versions of CACTI. Therefore, we are including Host ID 0 which only exists in those cases.

We are discouraging users from using this method since with CACTI 0.8.6x we are checking the status of the host prior to wasting time polling a downed host.

Bottom line is that your host count is therefore 1 higher than the total number of hosts. It's stupid, but true. I think I'll change it now.

TheWitness

DevilSun · Post by **DevilSun** » Thu Oct 07, 2004 5:11 pm

So did I do something wrong by setting up this Dell Switch as opposed to previously when you weren't checking for downed hosts?

I just upgraded to 0.8.6b BTW and giving it a few cycles...

What is the temporary work around, is it included in "b", and does it fix my problem or are we still trying to fix my problem? If it's any help, the switch worked prior to any 0.8.6 which is when you introduced polling only if the device is up I beleive.

Would it be possible for future versions to be able to turn off downed host detection?

Post by **TheWitness** » Thu Oct 07, 2004 7:59 pm

DevilSun,

Send me a complete debug log via e-mail at larryjadams@comcast.net. This does not make sense and I need to browse the code. Also, try using cactid for a few cycles and see if the devices becomes UP.

TheWitness

mondaka · Post by **mondaka** » Tue Aug 16, 2005 9:55 am

DevilSun · Post by **DevilSun** » Wed Aug 31, 2005 11:29 am

Nope, no solutions, any of these devices which were "Unknown" status can never be re-added and work. I've almost given up and want to start from scratch, although it'd be a super pain in the arse.

ichoudhury · Post by **ichoudhury** » Wed Aug 31, 2005 12:10 pm

I get that too with certain servers. I go to the server and restart SNMP service and that seems to resolve the issue.

Device Status "Unknown"...why???

Device Status "Unknown"...why???

Who is online