Smart-UPS RT 5000 XL Network spikes during SNMP monitoring

[BillP]

Hi,

We are experiencing random "high collision" events logged by our HP Procurve switches on the Smart-UPS RT 5000 XL ports while the UPS is being SNMP monitored - these take the normally idle port up to 90% utilization and are quite disruptive.
"About UPS" shows firmware version is "3" -
are there release notes for newer firmware that perhaps resolve this?

thanks

[BillP]

I received this notice, but I don't see it in the forum - so I'll post it and reply

You are watching the thread "Smart-UPS RT 5000 XL Network spikes during SNMP monitoring", which was updated on Aug 16, 2012 9:53:05 AM by Angela:

<Referenced content removed

For your convenience, the new message is included below:
--------------------------------------------------------------
we are not aware of an issue with this. the UPS itself probably has a network management card so we need to know the model of that - AP9619 or AP9631 most likely, and the firmware of that. Can you log in via the web UI and see under Administration->General->About and tell me what you have there.

it also might be helpful to indicate what port speed these are operating on and if you are using SNMPv1 or V3 and how often the UPS is being polled via SNMP and how many values at a time.

thanks
--------------------------------------------------------------

[BillP]

Ok, here is more information

The WEB UI responds very slowly and after 3-4 clicks STOPS responding at all.
I've tried 4-5 different UPSes - same result - none of them respond on the web UI anymore, although they respond to SNMP - so I pulled the sysDescr which looks like it has firmware versions:

SNMPv2-MIB::sysDescr.0 = STRING: APC Web/SNMP Management Card (MB:v3.9.2 PF:v3.7.3 PN:apc_hw02_a
os_373.bin AF1:v3.7.2 AN1:apc_hw02_sumx_372.bin MN:AP9619 HR:A10 SN: ZA0627006748 MD:06/28/2006)
(Embedded PowerNet SNMP Agent SW v2.2 compatible)

I also noticed just hitting the web UI seems to trigger the "high collision" events on the UPS switch ports - so its not random - we can reproduce it quite easily.

Please advise which firmware we could update

thanks

[BillP]

The switch sees these UPS ports operating at 100Mb - some Half Duplex, some Full Duplex.

[BillP]

So

apc_hw02_sumx_372.bin means the NMC is on v3.72
and MN:AP9619 is the model of the NMC?

I think 3.72 is the latest firmware for this network management card?

We might try hard coding the 100FullDuplex on the switch next (should we do it in ht the config.ini as well)?

After I ssh'ed into the NMC and rebooted, the Web UI was responsive again

thanks

[BillP]

we are not aware of an issue with this. the UPS itself probably has a network management card so we need to know the model of that - AP9619 or AP9631 most likely, and the firmware of that. Can you log in via the web UI and see under Administration->General->About and tell me what you have there.

it also might be helpful to indicate what port speed these are operating on and if you are using SNMPv1 or V3 and how often the UPS is being polled via SNMP and how many values at a time.

thanks

[BillP · Accepted solution]

these network management cards are older and yes, you have the latest firmware already.

are you using HTTPS, SSH, and is it SNMPv1 or SNMPv3? since these cards are older, they have a more difficult time supporting all of these secure connection methods at the same time and could bog it down.

yes, i would set the port speed specifically on the NMC eventually but i assume if its still set to auto, the switch will force it to 100M/full duplex anyway. might be good to see if there is a difference when you let the NMC still use auto and then force it.

i'd also be curious if these NMCs do this with a default configuration if you were able to reset one to defaults.

lastly, if using HTTPS, there was just a recent microsoft security update -> http://support.microsoft.com/kb/2661254 that blocks 768 key bit lengths. if you are using HTTPS and the NMC is using a self signed/generated certificate, it'd be 768 bits. this could prevent HTTPS access via the web UI if you were using windows and had this update applied. it was fresh in my mind so i figured i'd note it. the work around for that would be to use our security wizard tool to create a 1024 bit cert.