[
https://issues.apache.org/jira/browse/SOLR-13609?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16879399#comment-16879399
]
Erick Erickson commented on SOLR-13609:
---------------------------------------
Richard:
I usually recommend discussing this kind of thing on the user's (or dev list
if you're code-diving) before raising a JIRA, mostly because you get more eyes
on it faster.
That aside, are you sure you need to expungeDeletes? It's actually rarely
necessary. What i'm suggesting is that this may be useless work. However,
you've backed into a corner, see:
https://lucidworks.com/post/segment-merging-deleted-documents-optimize-may-bad/
That blog is about optimize, but expungeDeletes has the same issue of creating
potentially very large segments.
Much of this behavior has changed in Solr 7.5+, the link above links to another
blog about that.
All that aside, to answer your question: I don't know of a way to ask "what is
the current state of merging" simply. But you could take a snapshot of the
segments before you start by:
[http://solr:port/core/admin/segments|http://solrport/]
In your case you should see a bunch of segments disappear. Especially, given
the above, you should see one very large segment be replaced by another very
large segment when it's done. Admittedly, this is indirect evidence and it'd be
interesting if we could have an async optimize. And be a little careful, if
you're indexing at the same time you'll see segments come and go due to
background merging.
That said, since we discourage optimizing as much as we do, I doubt there'll
be a lot of interest in adding it unless you want to make a patch.
> Ability to know when an expunge has finished
> --------------------------------------------
>
> Key: SOLR-13609
> URL: https://issues.apache.org/jira/browse/SOLR-13609
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Affects Versions: 7.4
> Reporter: Richard
> Priority: Major
>
> At the company I work for, we do nightly expunges to clear down deleted docs
> _(providing the threshold is above 5% in our case)_.
> Whilst this has been okay for us, we want the ability to know when an expunge
> has completed. At the moment we do some calculations to estimate how long it
> would take.
> it would be nice if there was a way to see when an expunge has completed.
> This could either be by assigning an async id to the call, or any other means
> of having visibility.
> I started to look into this issue, but saw that the underlying call for
> expunging starts to use the lucene side of the code base, so thought I was
> digging to deep, so any advice on this issue would be much appreciated _(as
> I'm trying to contribute more to OOS)_.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
