Nuxeo 6 - nx_update_read_acls never ending process and High CPU usage. NX freezes
When we add new user rights to a folder Nuxeo freezes for 20 minutes and there is High CPU Usange for 20-30 minutes. Deadlock.
Last SQL sentences: org.nuxeo.ecm.core.storage.sql.jdbc.JDBCLogger SQL: INSERT INTO [acls] ([id], [pos], [name], [grant], [permission], [user], [group]) VALUES ('59753f95-7196-4235-aa0e-5ca55e7ba6ea', 3, 'local', false, 'Everything', 'Everyone', NULL) – BATCHED 2015-01-15 16:25:20,654 DEBUG [http-bio-0.0.0.0-8080-exec-80] [org.nuxeo.ecm.core.storage.sql.jdbc.JDBCMapper] updateReadAcls: updating 2015-01-15 16:25:20,654 TRACE [http-bio-0.0.0.0-8080-exec-80] org.nuxeo.ecm.core.storage.sql.jdbc.JDBCLogger SQL: EXEC dbo.nx_update_read_acls
After that EXEC dbo.nx_update_read_acls Nuxeo freezes, no more log activity but High CPU Usage, locks in the resources, etcetera, as you can see in the attached images.
I've checked it with SQL Profiler and it shows this:
Our Nuxeo database is not big, but this process reads millions of rows. Our hierarchy table has 28.000 rows only. The size of the directory where we're trying to add new rights to new users is around 500 Mbytes. If I do it in a 20-30 Mb folder there is no problems adding rights…
And querying some system tables I get this:
I've read that this problem was solved in earlier versions of Nuxeo. Is that right?
Do you have any fix for this?
Thank you, Gabriel
P.D.: last screenshot, identifying locks & blocks. It keeps for 20-30 minutes too.