First, I think 30ms is an incredible short time to...
# lucee
d
First, I think 30ms is an incredible short time to give up on a task. Lots of things could impact what might be considered normal execution time, such as a pause during garbage collection. If whatever task you're bringing is really completely handing the system after 30ms, I suspect you're not going to be able to safely terminate the thread. I don't know a lot about the innerads of how Lucee handles terminating threads, but I image much like timeouts work, if the thread is waiting on an external even to finishing, terminating the thread might not actually occur when you think it should. It could also leave things in an unstable state, causing memory leaks, etc.
e
30ms seek time on some SAN raid arrays. If it's a Windows Server not running on SSD, I would suggest defragging it. If its a vm on SSD, still defrag it, then clear the log files and restart your services. If it's still too slow, message me, there are a few more network tweaks to consider.
j
@dswitzer, the timeout is still to be determined
@dswitzer, we have some flukey problem in instantiating objects in mura's IOC guts on page requests and it bogs down. it's been very hard to troubleshoot since it's prod-only (can't reproduce it anywhere with load testing, etc.). you might be right that if there's some problem reading a file, or some sort of thing, it might not be killable. i still may try, tho.
@Evil Ware, the filesystem in question is internal to a docker container, which runs on an ssd-backed linux EC2 host. only small CFC files are in play.
e
You might want to consider that you have what is called a noisy neighbor in your ec2 environment, and you may not have swap configured on your ec2 host. Check out the OS internals first, what are the sys logs showing