high cpu load 2910al-48G

[Lutz Muehlig]

Hello,

we installed 3x 2910al-48G in our network with redundant links to the coreswitches.
From start we noticed periodicly raising cpu to 20% load on every switch (app. once a day).
Today on one switch cpu load was going up to 100% in one hour and then the switch was rebooting himself.
After reboot he works well, config was not lost.
Our logging server shows no warnings only the message: "System went down:..."

We use latest image W.14.38

Is this a known bug!?

Regards
Lutz

P.S. This thread has been moved from Switches, Hubs, Modems (Legacy ITRC forum) to ProCurve / ProVision-Based . -HP Forum Moderator

[Wim-Bart van der Waals_1]

Had the same here with 2 active 10GBe connections in a trunk (etherchannel). CPU did go rocked high, ports did go into blocking and finally the switch died.

We now have our 2910al switches connected as follow:

CORE 1 ---A1 2910al A2----A2 2910al A1--- CORE 2

Same principle except less performance when link between CORE and one of the 2910's is broken. But actually never seen our 10Gbe links exceed 4Gbps even on high load.

[Lutz Muehlig]

Thanks,

we have also the uplinks as trunks (2Gb) configured on the switch who reloaded yesterday.
trunk 47-48 Trk1 Trunk
trunk 45-46 Trk2 Trunk
The other 2910 is connected with 2x 1G and he had no problems till now.
Our cores are cisco.

But I don't understand how you changed your setup. What means A1/A2? Did you connect 2 2910 together?

regards
Lutz

[Gerhard Roets]

Hi Lutz

Mind posting the output of "show log -a" ?

Gerhard

[Lutz Muehlig]

Attached you'll find the log at the time of the reboot. There are no other events before/after the reboot.

Lutz

[Gerhard Roets]

Hi Lutz

The switch crashed with the following crash message.

M 07/20/10 15:44:43 sys: 'NMI event SW:PC=0x2d0001e5 LR:0xfffffee5 SP:0x9ff08cea Task='InetServer' Task ID=0x1a4f4000 '
I 01/01/90 01:00:06 00061 system: ----------------------------------------------
I 01/01/90 01:00:06 00063 system: System went down: 07/20/10 15:44:43
I 01/01/90 01:00:06 00064 system: NMI event SW:PC=0x2d0001e5 LR:0xfffffee5

I have not seen a fix for this in release notes. I would suggest you open a call with HP.

Which I could help more
Gerhard Roets

[cenk sasmaztin]

hi Lutz
please send me show tech all print your 2910 switch

[Lutz Muehlig]

Thanks but this is now not necessary.
I'm in touch with HP-support and got a new Image W.14.49, where the problem is fixed (hopefully).

regards
Lutz

[Lutz Muehlig]

Solution: new image