AEM instance disk size very bigger than it should be

[#1]

Hello!

On 5.6.1-SP2 AEM productive instances, I made data-packages to export them (with a total of a few tens of GB). After building the packages (on the author), the instance size increased on the disk.

Once the packages downloaded, I deleted them from the package manager. The reported instance size returned at its original size (before packages building), but the disk usage never decreased, and this that's a lot problematic...

The packages are probably kept by the versioning tool of AEM, even if they do not exist anymore in the package manager. But how to do to purge them ?

A very big thanks for your help, and merry Christmas and happy New Year to all of you!

Blaise

[RGBAnd · Accepted solution]

Dear RGBAnd,

Pretty good guide summarising 5.6.1 maintenance activities:

https://jcr-nosql.com/2015/02/24/aem-5-6-1-reducing-repository-size-and-nailing-down-web-site-naviga...

 Namely:

"

1.Purge workflow instance data /system/console/jmx and then choose com.adobe.granite.workflow “Maintenance” and choose purgeCompletedWorkflow.
2.Perform Tar Index merge it might take 10 to 20 minutes…please check error.log  choose repository under /system/console/jmx and tarIndexMerge.
3.Perform tar optimization manually that might go overnight under repository under /system/console/jmx by clicking “startTarOptimization“.
4.Perform Step 2 again that might go for 10 to 20 minutes again.
5.Finally run DataStoreGarbageCollection under repository /system/console/jmx… and finally you will see disk space used is reclaimed.

"

Regards,

Peter

[RGBAnd · Accepted solution]

Thanks a lot Peter, it works well ! I just tested on a local instance which has been reduced from 7.2 GB to 4.2 GB. I'm not testing this now (Friday 20th December at the end of the day) on the instances in production, which are several tens of GB, but I have a feeling that we'll be able to save a lot of disk space. Merry Christmas and Happy New Year!

[#4]

The problem is likely that although the package is deleted but still present in the datastore. A datastore GC is likely to reduce the disk consumption again.