You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ignite.apache.org by src <ka...@hotmail.com> on 2020/01/24 22:57:45 UTC

How to terminate long running transactions in 2.4.0?

Cluster details: 3 servers, 4 clients.

Ignite version: 2.4.0

Number of caches: 4

I am observing a lot of "Found long running transactions/cache futures" with
state=ROLLED_BACK for transactions on a particular cache which is causing
failure to start PME when any server node is booted. It looks like
txnTimeout was not set for transactions started on this particular cache.
Added the txnTimeout which will not cause any further "long running
transactions" but how do I eliminate the currently "long running
transactions" ? Reboot of client instances did not remove these
transactions.

1) Is there a way to terminate these transactions?

2) If I delete the particular cache will these transactions be terminated?

Thanks!

 



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to terminate long running transactions in 2.4.0?

Posted by Ilya Kasnacheev <il...@gmail.com>.
Hello!

VM-level deadlock is a deadlock on synchronized blocks in Java.

It's hard to say what happens in your case. Do you have a reproducer for
this behavior?

Regards,
-- 
Ilya Kasnacheev


пн, 16 мар. 2020 г. в 20:49, rc <ka...@hotmail.com>:

> Hi Ilya,
>
> Thanks for responding. Killing the originator nodes did not terminate the
> transactions. I would like understand more about the VM-level deadlock. How
> does one go about determining that?
>
> Thanks,
> rc
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>

Re: How to terminate long running transactions in 2.4.0?

Posted by rc <ka...@hotmail.com>.
Hi Ilya,

Thanks for responding. Killing the originator nodes did not terminate the
transactions. I would like understand more about the VM-level deadlock. How
does one go about determining that?

Thanks,
rc



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to terminate long running transactions in 2.4.0?

Posted by Ilya Kasnacheev <il...@gmail.com>.
Hello!

I think that killing originator nodes' of these transactions should
eventually cause them to terminate, unless there's a hard VM-level deadlock.

Regards,
-- 
Ilya Kasnacheev


сб, 25 янв. 2020 г. в 01:57, src <ka...@hotmail.com>:

> Cluster details: 3 servers, 4 clients.
>
> Ignite version: 2.4.0
>
> Number of caches: 4
>
> I am observing a lot of "Found long running transactions/cache futures"
> with
> state=ROLLED_BACK for transactions on a particular cache which is causing
> failure to start PME when any server node is booted. It looks like
> txnTimeout was not set for transactions started on this particular cache.
> Added the txnTimeout which will not cause any further "long running
> transactions" but how do I eliminate the currently "long running
> transactions" ? Reboot of client instances did not remove these
> transactions.
>
> 1) Is there a way to terminate these transactions?
>
> 2) If I delete the particular cache will these transactions be terminated?
>
> Thanks!
>
>
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>