[
https://issues.apache.org/jira/browse/OAK-3566?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14989641#comment-14989641
]
Julian Reschke commented on OAK-3566:
-------------------------------------
I agree that 100% reliability can not be achieved if the marker theoretically
could be gone before it's needed. The question is how reliable it needs to be,
and whether alternative approaches are really better.
A background read will *update* the cache. This is desirable and correct.
(That being said an update operation might remove the marker and reinsert a new
one, and that might create a timing window during which the problem might
surface again; but this can be fixed)
So the main issue appears to be "regular" cache eviction within the time window
of a long running query; are we really scared about that? (Note that we could
also add a simple protection making sure that slow queries never update the
cache)
> Stale documents in RDBDocumentStore cache
> -----------------------------------------
>
> Key: OAK-3566
> URL: https://issues.apache.org/jira/browse/OAK-3566
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: core, rdbmk
> Affects Versions: 1.0, 1.2
> Reporter: Marcel Reutegger
> Assignee: Julian Reschke
> Attachments: OAK-3566-test.patch, OAK-3566.patch
>
>
> This issue is about the same problem as described in OAK-1897 but for the
> RDBDocumentStore implementation.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
