Juniper stack polling - better but still hurting
Hello all,
Moving forward with our POC, and almost everything is perfect...the one thing that's still nagging us is the long polling times for our Juniper switch stacks. Our newest QFX 5k 96s 2-member stacks are now down under 300 seconds after upping max-requests to 70 (~230 seconds now), but our older EX stacks are still taking longer than the polling interval (~330 seconds) to complete.
I realize it's the length of time that the Junipers take to handle the snmpbulkwalk of IfEntry and IfEntryX, but has anyone come up with a workaround to date? Obviously stripping OIDs out of the standard MIBs is not a solution because it could break other devices, but is there a possibility I could create an "IfEntry-lite" version that would be exclusive to the EX platform? Or even a custom poller that uses JunoScript instead of snmp? I'm looking for any reasonable solution, even if it takes some work to implement.
Thanks!
Aaron
Hi Aaron,
I'm looking at ways of trying to alleviate these sorts of problems. One possiblity is moving some of the things we currently collect in the poller to another, less frequently run, process for instances like this. Things like Duplex, MTU, ifAlias, etc.
A major rewrite of the ports poller is somewhere on the horizon, I think.
adam. On 09/09/2015 18:12:46, Aaron Finney aaron.finney@openx.com wrote: Hello all,
Moving forward with our POC, and almost everything is perfect...the one thing that's still nagging us is the long polling times for our Juniper switch stacks. Our newest QFX 5k 96s 2-member stacks are now down under 300 seconds after upping max-requests to 70 (~230 seconds now), but our older EX stacks are still taking longer than the polling interval (~330 seconds) to complete.
I realize it's the length of time that the Junipers take to handle the snmpbulkwalk of IfEntry and IfEntryX, but has anyone come up with a workaround to date? Obviously stripping OIDs out of the standard MIBs is not a solution because it could break other devices, but is there a possibility I could create an "IfEntry-lite" version that would be exclusive to the EX platform? Or even a custom poller that uses JunoScript instead of snmp? I'm looking for any reasonable solution, even if it takes some work to implement.
Thanks!
Aaron
--
Aaron Finney | Network Engineer
888 East Walnut Street, 2nd Floor | Pasadena, CA 91101
office: +1 (626) 466-1141 x6035
Watch how we make online advertising simple: http://bit.ly/Ent_vid [http://bit.ly/Ent_vid] www.openx.com [http://www.openx.com/] | follow us on: Twitter [http://www.twitter.com/] Facebook [http://www.facebook.com/] LinkedIn [http://www.linkedin.com/]
_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Hi Adam,
I'm happy to hear it's something that's at least being considered, it would make a big difference for us. We are phasing out the older EX stacks anyway, but I'm sure there will be more than a few of them hanging around in some capacity for a long time to come. :D
Aaron
On Wed, Sep 9, 2015 at 5:36 PM, Adam Armstrong adama@memetic.org wrote:
Hi Aaron,
I'm looking at ways of trying to alleviate these sorts of problems. One possiblity is moving some of the things we currently collect in the poller to another, less frequently run, process for instances like this. Things like Duplex, MTU, ifAlias, etc.
A major rewrite of the ports poller is somewhere on the horizon, I think.
adam.
On 09/09/2015 18:12:46, Aaron Finney aaron.finney@openx.com wrote: Hello all,
Moving forward with our POC, and almost everything is perfect...the one thing that's still nagging us is the long polling times for our Juniper switch stacks. Our newest QFX 5k 96s 2-member stacks are now down under 300 seconds after upping max-requests to 70 (~230 seconds now), but our older EX stacks are still taking longer than the polling interval (~330 seconds) to complete.
I realize it's the length of time that the Junipers take to handle the snmpbulkwalk of IfEntry and IfEntryX, but has anyone come up with a workaround to date? Obviously stripping OIDs out of the standard MIBs is not a solution because it could break other devices, but is there a possibility I could create an "IfEntry-lite" version that would be exclusive to the EX platform? Or even a custom poller that uses JunoScript instead of snmp? I'm looking for any reasonable solution, even if it takes some work to implement.
Thanks!
Aaron
-- *Aaron Finney | **Network Engineer*
888 East Walnut Street, 2nd Floor | Pasadena, CA 91101 office: +1 (626) 466-1141 x6035
Watch how we make online advertising simple: http://bit.ly/Ent_vid www.openx.com | follow us on: Twitter http://www.twitter.com/ Facebook http://www.facebook.com/ LinkedIn http://www.linkedin.com/
_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
participants (2)
-
Aaron Finney -
Adam Armstrong