topic Re: Activiti transaction commit time in Alfresco Archive

Activiti transaction commit time

lmollea — Tue, 15 Apr 2014 10:10:51 GMT

Hi all.Has anyone experienced processes becoming stuck due to missed signals caused by Activiti engine write times? If yes, any suggestion on how to handle this situation at best?Scenario is this: we have process that is in a wait state (user task or signal catch handler). This process gets woken up

Re: Activiti transaction commit time

trademak — Wed, 16 Apr 2014 14:18:36 GMT

Okay. Do you have the history level of Activiti set to full? Lowering the history level would increase the performance of the Engine quite a bit. Is the database running a different server? So do you have network latency as well? Is there a specific query that takes the longest time?
It would also be an option to add a retry in the process instance signalling logic.

Best regards,

Re: Activiti transaction commit time

lmollea — Wed, 16 Apr 2014 15:58:35 GMT

History was set to default; lowered it to "activity" but there was no noticeable change in the number of queries issued in that transition: from 15 insert, 5 update, 8 delete, to 14 insert, 3 update, 8 delete.

Disabling history completely may not be feasible, we still plan on using the activiti history tables for monitoring purposes.

Databse tables are pretty empty, system is still under development and not released in production. Queries seems all performing similarly (2-3 ms each).
Think in the end all comes down to the number of network round-trips: Activiti does about 25 while the remote service invoked by the Service Task does (iirc) 2 queries on that call and there are then 2 more network round trips for the JMS message. So it's a 25 vs. 4 round trips, hard to beat…

Moving Activiti database locally (using embedded h2 for example) could probably make the issue highly improbable, but I don't know if it can work as we may need to cluster the environment in production.

Re: Activiti transaction commit time

jbarrez — Tue, 22 Apr 2014 10:46:12 GMT

> Moving Activiti database locally (using embedded h2 for example) could probably make the issue highly improbable, but I don't know if it can work as we may need to cluster the environment in production.

I wouldn't do that. H2 is not a production db.