What to do when we got the warn "Selection hierarchy.name ...is getting big and now has size" ?
Hi,
We have the warning “[org.nuxeo.ecm.core.storage.sql.Selection] Selection hierarchy.name for id=… is getting big and now has size: 15000” since few weeks.
What are we supposed to do with ?
How the threshold have been choosen ? Is it a per default value or calculated with server parameters ?
When I check in the database in hierarchy table with the mentionned id, primarytype is “TaskRoot”. psql=# select count(*) from hierarchy where parentid='..';
count
16410
All of them have for primarytype 'TaskDoc'. Are they all for unterminated workflows ?
Our users experiments troubles to use nuxeo (we had to stop/start it twice since yesterday as we got an error 500). Is this warning can have a relation with the difficulties and slowless observe ?
Thank you for your help
Note : We have a nuxeo 5.8.0-HF31
Hi,
This is a warning because you have more than 15k documents in a container and this might reveal a performance problem. You can get rid of this warning by using the org.nuxeo.vcs.selection.warn.threshold property using a bigger value.
But in your case may be your problem is related to this long list of workflow tasks, you need to check the state of theses tasks to see if they are unterminated or if there is a cleanup problem, see https://jira.nuxeo.com/browse/NXP-12621 and https://jira.nuxeo.com/browse/NXP-15085.
Regards
ben
Thank for your answer. However, how to determine if a task is terminated or not ?
I've checked the task_info table :
psql=# select distinct(ended), count(*) from task_info group by ended ;
'f' : 735
't' : 2773
(and taskdoc number had increase since yesterday to 16505)
I've try this (but I don't now if that request had the mean I'm looking for or it's a non-sens):
psql=# select count(id) from hierarchy where parentid='…' and id in (select taskdocid from task_info where ended='t'); -> 2773
Does it means that all ended tasks are no clean up ?
(comment of the 9 july 2015 - wrongly in answer first)