Mailing List Archive: XMR slow snmp

Hi!

I’m trying to monitor a few foundry XMR’s via SNMP, however I’m experiencing issues where the XMR is VERY slow at responding, if it responds at all. I’ve confirmed it’s not an ACL issue (there’s no receive ACL, etc) and the CPU while a little busy, is not crazy. (40%usage overall, SNMP only tacking 1% average). snmpget’s timeout 3/4 of the time. snmpwalk’s are impossible, as even if a few values are returned, it will inevitably time-out before it reaches the end.

I must admit I’m a bit new to the XMR platform, can anyone let me know what I could check regarding this?

Here is the current snmp configuration:

snmp-server
snmp-server community 0 OUR_COMMUNITY ro 10
snmp-server max-ifindex-per-module 64
snmp-server cache size 200
snmp-server cpu max-non-idle-utilization 25

My IP is of course in access-list 10.

Not super familiar with how the cache works, but it shows the following currently:

SNMP cache statistics:

Configured size: 200KB
Configured maximum age: 500ms
Percent of space in use: 0%
Total objects requested: 195679182
Total object prefetches requested: 3310250001
Total requested objects found in cache: 173081906
Total objects aged out: 72533169
Total objects removed for space: 0

Cache hit rate: 88%
Percent of objects aged out that were unused: 9%
Percent of objects removed for space that were unused: 0%

Thanks!
Thomas

Do you have any idea how many SNMP requests/sec you're sending to the
device in question? I've seen timeouts like this when the device is
overwhelmed with SNMP requests. Perhaps you have another NMS system that is
sending a lot of queries, or this NMS system is sending them too rapidly?

As far as an snmpbulkwalk, this should be perfectly functional on an XMR
that isn't receiving a lot of SNMP contention from other NMS systems.

-- Eric Cables

On Fri, Feb 27, 2015 at 1:49 PM, Thomas St-Pierre <tstpierre@iweb.com>
wrote:

> Hi!
>
> Iâ€™m trying to monitor a few foundry XMRâ€™s via SNMP, however Iâ€™m
> experiencing issues where the XMR is VERY slow at responding, if it
> responds at all. Iâ€™ve confirmed itâ€™s not an ACL issue (thereâ€™s no receive
> ACL, etc) and the CPU while a little busy, is not crazy. (40%usage overall,
> SNMP only tacking 1% average). snmpgetâ€™s timeout 3/4 of the time.
> snmpwalkâ€™s are impossible, as even if a few values are returned, it will
> inevitably time-out before it reaches the end.
>
> I must admit Iâ€™m a bit new to the XMR platform, can anyone let me know
> what I could check regarding this?
>
> Here is the current snmp configuration:
>
> snmp-server
> snmp-server community 0 OUR_COMMUNITY ro 10
> snmp-server max-ifindex-per-module 64
> snmp-server cache size 200
> snmp-server cpu max-non-idle-utilization 25
>
> My IP is of course in access-list 10.
>
> Not super familiar with how the cache works, but it shows the following
> currently:
>
> SNMP cache statistics:
>
> Configured size: 200KB
> Configured maximum age: 500ms
> Percent of space in use: 0%
> Total objects requested: 195679182
> Total object prefetches requested: 3310250001
> Total requested objects found in cache: 173081906
> Total objects aged out: 72533169
> Total objects removed for space: 0
>
> Cache hit rate: 88%
> Percent of objects aged out that were unused: 9%
> Percent of objects removed for space that were unused: 0%
>
>
> Thanks!
> Thomas
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp@puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>