keto memory leak. im using v0.11.1 not sure if its...
# talk-keto
m
keto memory leak. im using v0.11.1 not sure if its just my configuration yet though. anyone experiencing the same? (down spikes are from keto pod restarts)
Hi @bland-eye-99092, really sorry for the tag -- I don't know whos responsible for keto. Just really want to get insights about this. Hopefully you can help point me to the right person 🙏 Just fyi, we're using keto on production for our apikey authorizations. We do about 10 keto rule checks per second. The above image is directly from our GCP monitoring. I initially thought it was due to our outdated keto version, but after updating to the latest, its still happening.
c
I mean, this may not be the answer you want, but did you consider simply forking keto and running some pprof'ing?
Another hot tip is to send your running keto instance a SIGABRT to kill it (of course not during the wildest prod hours). That will dump all running threads (goroutines) to stdout, allowing you to weed out any potential contenders for deadlocks etc.