We have been having some stability issues with our F5 after upgrading them to 11.2. We had no such issues in the lab but for some reason production does not behave exactly the same as the lab no matter how much we try to make it the same.
F5 support was baffled and even replaced our unstable unit. EUD report was clean so replacing the unit was only done because of the lack of other explanation.
After a day with the replacement unit the same unstable behavior started. Reboots every 15 minutes or so.
A member of my team figured out that the unit syslog-ng was having an issue with the tcp connection to a remote server. Syslog-ng would lose the connection and start consuming all the CPU resources. Eventually the F5 OS decides that it is time to reboot because other processes can’t complete their task.
We have since migrated the sysloging to UDP which shows no signs of the issue.
F5 support is asking us to test a fix where they have upgraded syslog-ng from 2.0.8 to 2.1.4. This fix is based on 11.2.0 HF1 which is a bit odd since HF2 has been out for a little while.
