4
responses
responses
We have an application that does some multiple updates every morning between 6am and 7am. During this time, it generates massive amounts of log entries.
This in turn causes the box to run out of memory, triggering Linux's OOM daemon. Running the NxLog-ce.
I have added
PersistLogqueue TRUE
SyncLogqueue TRUE
CacheFlushInterval always
CacheSync TRUE
To the config, and will see if that makes a difference, but it seems that is more to safeguard messages from being lost.
I have looked at https://nxlog.co/question/802/nxlog-ce-memory-leak and https://nxlog.co/question/4132/cache-disk-works-not-good, but not sure those two are what I am after?
Comments (4)
Hey Werner,
Not sure if the same case. Can you paste your config here?
Hi,
Can do - it seems to be a memory leak: https://postimg.cc/f3jyP17z For now, I have the service restarting automatically at 1pm, which effectively resolves the issue, but I would like to properly solve it.
I would try to make the config simpler in order to isolate the line causing the memory leak.
Unfortunately the feed forms part of the monitoring strategy, so I will not be able to isolate the offending input. If I start adding one at a time back, it means that the remaining input will have to be offline for days as I'd image I'd have to give each input a 24h window to make sure all the race conditions are met.