PCM+ B_02_10_5 bogus traffic

[Les Ligetfalvy]

On July 3rd, my PCM received autoupdate package #5 and since then I get bogus high water marks. I called NetHelp and was told to remove all traffic settings, delete the switches, let them rediscover, and add back all the traffic devices. I've done this before and was pretty sure I got the security settings all done globally but not having much joy now. Now the traffic monitor reports all segments as "not reporting" and there are security related errors in the event log.

It should not be this difficult. I can use the CLI Wizard and it takes the defaults just fine but but if I try to add Authorized Managers using Device Manager, it fails.

Hmmm... time for another phonecall I guess...

[Jonathan Axford]

Keep us informed on what they say... Would be interested in finding out if this is a general bug with Update 5 or whether it is soecific to your kit....

Cheers!

[Les Ligetfalvy]

Well... I decided to hold off phoning just yet and worked through the security errors. For some reason, the write community string changed in the global preferences. All the other switches that I have in PCM without traffic monitoring, I did not delete/rediscover and they were all set to "Use PCM Defaults" too yet they worked. What finally caught my eye was the number of asterisks masking the write community string didn't match the number of chars in the string. Go figure!

So, the Test Communication Parms wizard passes now with flying colours but still no devices reporting. Do I delete all the traffic settings now and start over or sit on my hands and wait?

[Les Ligetfalvy]

OK, got the circulation going in my hands again... :p

Removed the traffic monitor settings and added them back in and now the segments are all responding. Time will tell if the bogus values are gone, never to return... so far so good.

Oh well, I had a few ports that were discovered at the wrong speed previously so rebooting the switches and rediscovering them sorted that at least. I can't wait for the the next release that purports to fix this speed thing.

Dang! whilst typing this out, I glanced up at my other monitor and see a bunch of pegged needles and high water marks. See attached. Time to call NetHelp I guess.

Curious whether anyone else is seeing this with update 5?

[Les Ligetfalvy]

Escalated to division.

[Per Larsson]

I can say that I have the exact same problem. Just updated to update 5 and the traffic monitor goes bananas! Even on ports not connected. But only on the subnet that the PCM is on, not on any of our other subnets.

[Mike Satterfield]

Add me to the list too! Same symptoms even after re-discovery, re-establishing "stats" and "sampling" settings, and a server restart.

I don't see a relationship with the subnet proximity to the PCM server however. Only a select portion of my network switches are included in the free-for-all.

Mike

[Les Ligetfalvy]

FYI my incident number is 3215393241 in case you want to call in and be added to the list.

[Les Ligetfalvy]

A division engineer was able to duplicate this issue in his test lab late yesterday. No timeframe for a fix and the only proposed work-around is to completely un-install and re-install PCM with Update 4.

I cannot find any upadates for PCM+ 2.1 at http://www.hp.com/rnd/software/network_management.htm although they are still available through the FTP site.

I will it as-is and fly blind for now and see what Monday brings.

[Les Ligetfalvy]

Oh, I forgot to mention that I did NOT enable the setting "Download and install automatically", opting only for "Notify if updates are available" but yet somehow automagically this update installed itself. :(

I don't want to make a big stink about it because I would have installed it anyway if notified and would have ended up in the same situation I am in now. What does bother me a little though is that I was unaware of the automagic update and based partly on the bogus reporting, I scheduled an emergency service window to reboot the entire mesh. It was due for a reboot anyway as one of the switches was ignoring too many pings and SNMP gets and WebAgent on another was not responding.