Hello!

Can you provide stack trace from that thread?

No, this is a different thread than exchange worker.

I don't think that cluster is supposed to fail clients in this case. We
have timeout for transactions waiting on PME now.

Regards,
-- 
Ilya Kasnacheev


сб, 12 дек. 2020 г. в 10:23, Gangaiah Gundeboina <[email protected]>:

> Hi Ilya,
>
> Again we got same issue 'tcp-comm-worker-#1' blocked. We have observed one
> thing whenever multiple clients joining to cluster at same time, the error
> is printing. I could see there is only one thread(tcp-comm-worker), checked
> from logs and also from visualvm.
>
> From documentation also, could see there is only one thread per node below
> are details. Is it same thread?
>
> #######################################################
> Initialization of exchange means adding future
> GridDhtPartitionsExchangeFuture to queue on each node.
>
> Put to this queue (GridCachePartitionExchangeManager.ExchangeWorker#futQ)
> is
> done from the discovery thread.
>
> 'Exchange worker' thread manages this queue. Using only one exchange worker
> thread per node provides a strict processing order for futures.
> ########################################################
>
> And also I heard cluster will kick the clients where more
> transactions/atomic updates happening while waiting for partition-exchange
> to complete. Is it could be reason, could you please help us to find out
> the
> cause and fix the issue.
>
> Thanks and Regards,
> Gangaiah
>
>
>
> -----
> Thanks and Regards,
> Gangaiah
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>

Reply via email to