[oak-segment-tar] Question about CompactionAndCleanupIT

[Andrei Dulceanu <du...@adobe.com>]

Hi,

I am ramping up on compaction/cleanup stuff in oak-segment-tar and while reading the code in CompactionAndCleanupIT, found this line [1]. Now my question is this:  why does it take two compaction/cleanup cycles to remove the segment(s) storing that blob?

Regards,
Andrei

[1]   https://github.com/apache/jackrabbit-oak/blob/trunk/oak-segment-tar/src/test/java/org/apache/jackrabbit/oak/segment/CompactionAndCleanupIT.java#L139

Re: [oak-segment-tar] Question about CompactionAndCleanupIT

[Michael Dürig <md...@apache.org>]

Hi,

With the changes done mainly with OAK-3348 the cleanup phase is now 
based on generations: each compaction would create a new generation 
completely independent on the previous one. Cleanup would remove 
everything from older generations respecting a retention time of by 
default 2 generations. Using a retention time of 1 would most certainly 
lead to SegmentNotFoundExceptions as segments created seconds before 
compaction finished would already be eligible for cleanup.
See my comment [1] explaining the patch with the initial fix for 
OAK-3348 for greater detail.

https://issues.apache.org/jira/browse/OAK-3348?focusedCommentId=15190912&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15190912

See

Michael

On 10.8.16 11:10 , Andrei Dulceanu wrote:
> Hi,
>
> I am ramping up on compaction/cleanup stuff in oak-segment-tar and while reading the code in CompactionAndCleanupIT, found this line [1]. Now my question is this:  why does it take two compaction/cleanup cycles to remove the segment(s) storing that blob?
>
> Regards,
> Andrei
>
> [1]   https://github.com/apache/jackrabbit-oak/blob/trunk/oak-segment-tar/src/test/java/org/apache/jackrabbit/oak/segment/CompactionAndCleanupIT.java#L139
>