hi there,
I found this case when using BPT in java stack in production environment:
# the bpt stuck at IF activity for some bpt instances (about 3 instances out of thousand) - waiting state
# amazingly, after one day, it runs smoothly
# in the error log, there is messages something like: activity does not exists
# my question is what cause this waiting?
## my analysis is here, but not sure:
### there is lack of resources
### then outsystems swap the activity from memory to disk
### and then swap back to memory to rerun again
### it gets back CPU time after one day (is there internal algorithm /scheduler for this)
best regards,
brrr
Hi "this",
It's difficult to sat what caused the instances to "hang" like that, but the fact that they finally moved on indicates to me that either there were just a lot of things to process, and these instances came at the back of the queue, or their conditions actually didn't evaluate to anything that allowed them to progress. Either way, I'd say keep an eye on things, and if this happens more often, better investigate what causes this.
That said, your "analysis" is really invalid. The Platform runs on IIS, which runs on Windows Server. Any kind of memory-to-disk swapping is done by the OS, not even by IIS, let alone by the Platform. There's no way that an application running on IIS can actively "swap" anything (though it is possible that an application uses too many resources, and IIS recycles a pool, but that's a different thing).
# Nobody experience this?