Moving the discussion back to the devlist. Hi Anjil,
The numbering of the thread names is expected to increase, there's no issue with that. The fact that you're reporting a lot of threads being active based on your thread dump is definitely something to take care of. Probably at some places in Fineract the executors should be switched to be thread-pool based executors instead of simple ones. Can you pls file a ticket for this? Thanks. Best, Arnold On Tue, May 30, 2023 at 9:34 AM Anjil R Chinnapatlolla <anchi...@in.ibm.com> wrote: > Hi All, > > > > There seems to be some issue with my below mail delivery to the community. > > Can you please review my below observation sent to community mailing list > earlier and share your thoughts. > > > > Thanks & Regards, > > Anjil > > > > *From: *Anjil R Chinnapatlolla <anchi...@in.ibm.com> > *Date: *Thursday, 18 May 2023 at 1:51 PM > *To: *dev@fineract.apache.org <dev@fineract.apache.org> > *Subject: *[EXTERNAL] Too many threads of type "SimpleAsyncTaskExecutor-" > accumulating > > Dear Dev community team, I got to observe too many threads of type > “SimpleAsyncTaskExecutor-86046” getting accumulated while running Fineract > with increased concurrent users (ranging from 200-1000 users at a time > performing saving deposits > > ZjQcmQRYFpfptBannerStart > > *This Message Is From an External Sender * > > This message came from outside your organization. > > ZjQcmQRYFpfptBannerEnd > > Dear Dev community team, > > > > I got to observe too many threads of type “*SimpleAsyncTaskExecutor-86046*” > getting accumulated while running Fineract with increased concurrent users > (ranging from 200-1000 users at a time performing saving deposits across 5 > pod replication). > > > > When deployed in a Kubernetes environment with read/write pods separated, > the accumulation of threads within the pods is causing the resource > exhaustion and JVM freeze. > > > > From a quick java thread-dump walk-through, my understanding is that these > threads are created in the process of business event notification post > write commands (like savings deposit) and is required by design. Raising > this observation to see if it can be optimized anyway to reduce the number > of threads getting generated per transaction. > > > > From one of the thread dumps of a pod, we observed 168 threads of type “ > *https-jsse-nio2-8443-exec-*“ and “*SimpleAsyncTaskExecutor-*“ were 320. > > > > Thanks & Regards, > > Anjil >
