CPU spike on SRX650, causing TCP sessions to pile up and drop.
I have a very annoying problem here where randomly, the CPU/SPU on our SRX650 cluster (active node only) will spike up to 99% and TCP sessions will pile up and get dropped. The session spike does appear tp happen a minute or so before the CPU spike The issue will magically correct itself one to two minutes later and traffic will begin flowing normally again. There is not a corresponding spike in bandwidth, either inbound or outbound. I have been over the systems with support backwards and forwards and nothing has come up. The only thing I can find that is close is this PR, but that was apparently resolved in 12.1X44-D25 and we are running 12.1X44-D40.2. Our firewalls have a simple setup with three zones - trust, untrust, and dmz, with traffic hitting our load balancers in the DMZ and servers in the trust zone occasionally talking to servers on the internet via NAT on the SRX cluster. Does anyone have any suggestions on what to look for as a cause of the TCP session spike?