Cacti 0.8.7 cmd.php to spine, several hosts show down.

nvetro · Post by **nvetro** » Tue Dec 18, 2007 11:35 am

Hey all, I upgraded from cmd.php to spine. It appears spine is working great, it finishes ALOT fast then cmd.php. Now onto my issues:

Out of my 50 hosts, about 15 of them shown "down" there is NOTHING in the log with regard to these hosts as they show "down" and will not even be considered to be polled when the cron job runs. When I click on the "down" device the snmp info shows fine. The hosts that show "UP" are polling fine and everything works fine. The hsot detection method is set to "SNMP" only and the snmp time out is 500 (default)

What do you guys think?

nvetro · Post by **nvetro** » Tue Dec 18, 2007 2:49 pm

Actually i missed the error, here the the error I get which what is causing my those few hosts to be down:

12/18/2007 02:45:57 PM - SPINE: Poller[0] Host[62] Hostname[192.168.0.5] ERROR: HOST EVENT: Host is DOWN Message: Host did not respond to SNMP

Mind you, the SNMP info banner displays fine for this device, and this device also shows up and polls fine when cmd.php is set for the poller type. What do you guys think?

Again, I am running Cacti 0.8.7 with spine 0.8.7a (I don't think upgrading will solve the issue) It has to be something silly with those few hosts, as the other hosts are polling perfect. When I compared the up hosts to the down hosts all the settings were the same, what do you guys think?

nvetro · Post by **nvetro** » Tue Dec 18, 2007 3:40 pm

Update: If I create a host, she stays down, and also doesn't create the graph files.

Something silly with SNMP, any ideas?

Post by **TheWitness** » Tue Dec 18, 2007 6:22 pm

Well, from the web site can you walk the hosts?

TheWitness

fmangeant · Post by **fmangeant** » Wed Dec 19, 2007 3:26 am

Hi

what "Downed Host Detection" method are you using ?

nvetro · Post by **nvetro** » Thu Dec 20, 2007 3:08 pm

fmangeant: the default, SNMP only.

Thewitness: how do i walk the hosts from the website?

Post by **TheWitness** » Thu Dec 20, 2007 10:25 pm

ssh into the box and then "snmpwalk -c <your_community> -v <your_version> <your_host> .1.3

TheWitness

nvetro · Post by **nvetro** » Tue Feb 05, 2008 10:23 pm

Update: Still not able to get Spine to work (cmd.php) works fine though. I upgraded to cacti 0.8.7a and also went as far as moving it to a brand new box.....when is switch to spine, errors spew out about snmp time outs and cacti downs about 20 of my 50 hosts (cmd.php) does not do this...when i snmpwalk a downed host it runs fine..for alittle bit (about 10 full seconds)...then times out around the same location...I am running the snmpwalk mib as states above, i'm doing, the hosts that are up i run the same snmpwalk to and it finishes fine:

/usr/local/bin/snmpwalk -c cacti -u cacti -A <passwprd>-v 3 -l authNoPriv <ipaddress> .1.3

Anyone have any ideas? For the record, while using spine, the hosts that are not downed collect data and graph fine using spine...if i use cmd.php they are all fine and poll......What gives guys?

nvetro · Post by **nvetro** » Tue Feb 05, 2008 10:33 pm

Here is the result of an snmpwalk on a host that shows down in cacti (using spine), it runs for a while, then just times out at the same spot...same with the other downed hosts. What do you all think?

Post by **TheWitness** » Wed Feb 06, 2008 6:19 am

Can you reduce the MaxOID count in the poller setup and see if this has any affect.

Larry

nvetro · Post by **nvetro** » Wed Feb 06, 2008 9:02 am

I set the "The Maximum SNMP OID's Per SNMP Get Request" which is in Configuration>Settings>Poller tab to 10....

Did you mean this or the MAX oid setting in each host device? That is also set to 10 by the way....

Attached is a screen shot of my poller tab settings. Don't think this will work as I had this set to 60, and tried it from the default which i think was 10.

nvetro · Post by **nvetro** » Wed Feb 06, 2008 9:36 am

For good messure, i have attached the top portion of my technical support page:

nvetro · Post by **nvetro** » Wed Feb 06, 2008 9:41 am

Also i'm using the newest Cacti 0.8.7a and Spine 0.8.7b

# ./spine -v
SPINE 0.8.7b Copyright 2002-2007 by The Cacti Group

nvetro · Post by **nvetro** » Wed Feb 06, 2008 10:50 am

Doesn't look like changing the MAX OID did anything, some of the hosts are still showing down. What other information can i provide that will assist us?

Post by **gandalf** » Wed Feb 06, 2008 12:37 pm

Please redics concurrent processes to 1-2 times the number of cores of your server
Reinhard

Cacti 0.8.7 cmd.php to spine, several hosts show down.

Cacti 0.8.7 cmd.php to spine, several hosts show down.

Who is online