iLO3 becomes unusably slow

[Bruce Edge]

If you are logged into the iLO ssh connection, and it sits idle for a while, it'll start taking 5-20 sec to echo each typed char back. It takes a hard power removal (plug) to get it to reset.

This is happening on Proliant dl380 G7's.

I've spent the better part of a week on HP chat support on this and had it escalated and am still waiting for a callback.

I've upgraded iLO FW, BIOS FW and HP even send a new MOBO. Problem remains.

a) Is anyone else seeing this?
b) Is there a solution?

Thanks

-Bruce

[Bryan Gartner]

Can you provide some more detail and answers to the following:

1. How long is the time before the ssh slows down?

2. After logging into the SSH of iLO (e.g. the SMASH CLP), were any commands executed?

3. Is there "session timeout" established or simply using the defaults.

4. Are you using the Virtual Serial Port (VSP) feature, this would be of interest.

[Michael Leu]

Maybe this issue is related?
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1449710

It really bothers me that as soon as iLO2 has become mature/stable, HP had to throw it all away and let the buggyness start all over again with iLO3.

[Bruce Edge]

It's not related the the ssh timeout because when it happens, a new login takes a long time to issue a prompt and one has to type the passwd with 10 second pauses after each character.

Although, it could have been a previous login that timed out and triggered this.

The only relevant setting that I can see are:
Idle connection timeout: infinite

It would be nice to have telnet support again, if only for conserver management of the VSP port.

It happens after the system has been up for a while. Seems to happen more often when I leave an ssh session connected.

I do use the VSP extensively, but it doesn't seem related as it goes into "slow mode" with or without the VSP active. The serial console is slow too when this happens.

Note this never happened with iLO2.

[CD9]

My guess is that something incoming on the iLo ethernet is the cause of this.

I noticed a similar problem using the ilo BIOS setup (F8 during boot) on my old iLo2, and pulling the ethernet cable fixed it.

There's updates available that mention in vague terms some kinds of TCP traffic killing the iLo card - maybe they've not fixed the bug for *all* kinds of traffic?