anyone running 5.4.0.80 or 5.4.1.8 in prod?
# lucee
z
anyone running 5.4.0.80 or 5.4.1.8 in prod?
c
Yes, 5.4.0.80. Been solid for over a fortnight.
🎉 1
m
5.4.1.8 - no issues so far
🎉 1
l
5.4.1.8 and all is good so far
j
we tried 5.3.10.120 then 5.4.0.80 but had to rollback to 5.3.10.97 because of random requests running very slowly for no explainable reason
z
Anything show up in debugging or with fusion reactor?
j
We spent a bunch of time trying to track down the slow requests in fusionreactor, but they seem to happen very infrequently/randomly we never nailed down a single source. It sounded like LDEV-4405 where a request that would normally take <1s would take >30s so we updated to 5.4.0.80 but the issue persisted. Not sure its helpful but we have a simple getTickCount() timer starting from the top of onRequestStart() and returned as a header in onRequestEnd() and even though the request was taking >30s to come back to the client the header would still say <1s, so the slow down was happening either before the queued request was starting to process or after it had completed.
These are on ECS instances with 4cpu and 8gb of ram, not high request loads, I thought maybe garbage collection, but the heap is getting 6gb of ram and its never using more than 3gb, doesn't seem to be running out of memory
j
5.4.0.80 here. (waiting for the jdk 5.4.1.8 docker image before i can use the latest.)