High Memory Usage of Control Tier Server

[Dan Nemec]

We're running Control Tier 3.6.1 and have a busy server that has projects for several QA environments. They get new packages frequently and we purge the objects in each project often, but it seems some vestiges hang around and are now causing Jetty to consume a lot of memory.The Jetty heap will consume 2.5G immediately on startup and will after short use become very slow.

Is there a way to purge some of the old data without removing the project in such a way that it will reduce the memory of jetty?

Will removing the project and creating again be totally clean if they have the same name?

Are there any maintenance best practices for keeping a Control Tier server lean and mean over time?

Here is some output from our server. The heap dump is over 300M. It loads about 2.5GB immediately into heap, and this is immediately promoted to the old generation where it stays there for the life of the VM.

Class Name Instances Size
char[] 7255226 (26.1%) 679448414 (36.8%)
java.lang.Object[] 1672509 (6%) 342370368 (18.5%)
java.lang.String 7237012 (26%) 260532432 (14.1%)
java.sql.Timestamp 3097162 (11.1%) 111497832 (6%)
org.hsqldb.MemoryNode 1588472 (5.7%) 95308320 (5.2%)
org.hsqldb.Row 1570782 (5.7%) 62831280 (3.4%)

Thanks,
Dan

[Charles Scott]

Hi Dan,

Have you tried compacting the object/type model and the map?

There's info here:  http://doc36.controltier.org/wiki/Workbench_RDF/XML_file_maintenance

that discuss model administration that may help trim it down significantly.

--
You received this message because you are subscribed to the Google Groups "ControlTier" group.
To post to this group, send email to contr...@googlegroups.com
To unsubscribe from this group, send email to controltier...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/controltier?hl=en
http://wiki.controltier.org