NGX Primary HA active node hard lock

[l0wkey]

Hello, here is what we are seeing. We are running a HA cluster in active / passive and the active node will randomonly hang after a day or so. The passive picks up and becomes active with no issues. When the system locks it's a complete hard lock and you cannot even access the console. We then moved the passive node to active and it also locks up randomly and then the passieve (the old active) becomes active and he will eventually die as well. Here is our specs


Checkpoint NGX HFA04 running on SPLAT
SecureXL
Intel Quad Gig cards
HP DL380 servers

Any info would be appreciated

[Izzio]

I have had a similar experience with R61, SPLAT, SecureXL, HP380 and Intel Quad GT cards.
The primary member, after that the secondary was stopped for maintenance, is crashed (it runs about 48h) and the secondary is suddenly crashed three hours later.
Now after a sleepless night I have made a rollback to R55 on a Crossbeam platform. :-((

There is a CP article describing compability problems between secureXL and Broadcom NICs (2x on the MB of HP380), but only regarding R55 ...I think that this article must be updated!

[l0wkey]

Wow not fun man. Weird thing is we have this same configuration deployed in a few diffrent places, only seems to be having issues in one cluster.

[l0wkey]

As a update, we have moved the cluster to active / active in unicast mode. We still have the same lockups. One system will lock the other is fine, then after about 72 hours the other system will lock and the other will be fine, It just keeps round robin locking.

[Robby Cauwerts]

Did you search in /var/log/messages* for usefull information?

[l0wkey]

Originally Posted by Robby Cauwerts

Did you search in /var/log/messages* for usefull information?

Sure have it yields NOTHING. Checkpoint wants us to enable kernel debugging, but last time we did that, it caused alot more problems and did not yield any more information.