in our environment, the logmanager seem to be completely screwed up (access.log is rotated every 10min instead daily, update.log is never rotated instead daily, file delete does not occure etc).
Is there a possibility to restart the logmanager without restarting MWG?
the Log Manager is not a Daemon but called via Cron every coueple of minutes. If you stop Cron it won´t be executed anymore. Maybe you want to stop Cron and have a look at what is happening?
I have never heard about such issues, but it sounds strange. Maybe you can file a SR with Support?
thanks for the reply.
There is another "funny" thing with it: Upon log file pushing, the HA secondary correctly appends the IP address to the log files, while on the HA master, the MAC address is added (or the system UID....). Obviously, the mwg-logmanager.rb script has some problems detecting the IP address if there's a VRRP address.
Is this a bug?
Okay that should work usually. I am not aware of an issue here - we need to have a look into both problems I think. Can you file an SR so that we can have a deeper look at this?
Oops, my fault.....
The IPv6 settings for the master HA were set to DHCP instead of Disabled. So the logmanager discovered an IPv6 address and used it......
If the logs are rotating too frequently, it might be because it rotates on a file size instead of timed basis. Are the logs regularly hitting the maximum file size define in log file?
No, they are not rotated due to size, but on a time basis only (daily).
We did some further investigation and found some communication problems to the FTP server.
It seems that if there are some problems accessing the ftp server for a large number of files, and the logmanager does not get through until next cron trigger, it get's messed up somehow.
We changed to another ftp server where the files are transfered within a few seconds and things look much better now.