[
https://issues.apache.org/jira/browse/TIKA-4100?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17744717#comment-17744717
]
Tim Allison edited comment on TIKA-4100 at 7/19/23 5:17 PM:
------------------------------------------------------------
I tried adding more indices. It turns out that we were already indexing the
mime_id columns, so that didn't help. The only thing that helped was creating
literal tmp tables and then doing a simpler update from these already computed
tmp tables.
I'm less than thrilled with adding more steps to the sql calls, but it at least
works now. I don't have an explanation of why we haven't hit this before, but
the ~infinite loop was
was (Author: [email protected]):
I tried adding more indices. It turns out that we were already indexing the
mime_id columns, so that didn't help. The only thing that helped was creating
literal tmp tables and then doing a simpler update on the already computed
table.
I'm less than thrilled with adding more steps to the sql calls, but it at least
works now. I don't have an explanation of why we haven't hit this before, but
the ~infinite loop was
> Work-around for ~infinite loop in one of the reports in tika-eval Report tool
> -----------------------------------------------------------------------------
>
> Key: TIKA-4100
> URL: https://issues.apache.org/jira/browse/TIKA-4100
> Project: Tika
> Issue Type: Bug
> Reporter: Tim Allison
> Priority: Minor
>
> For some reason, this query is now hanging forever in 2.6.0-2.8.0 on the new
> regression set of files.
>
> {noformat}
> update exceptions_compared ec set
> exc_cnt_a = (
> select count(1) as cnt
> from exceptions_a ea
> join profiles_a pa on ea.id=pa.id
> join profiles_b pb on pb.id=pa.id
> join mimes ma on pa.mime_id=ma.mime_id
> join mimes mb on pb.mime_id=mb.mime_id
> where ma.mime_id= ec.mime_id_a and mb.mime_id=ec.mime_id_b
> group by ma.mime_id, mb.mime_id); {noformat}
> The inner query doesn't take very long... I'm not sure why this is hanging,
> but it is a showstopper.
>
> I'm going to experiment with creating indices on the exceptions_compared and
> the related tables/fields.
>
> If that doesn't work, we may need to create temp tables and join them. :P
--
This message was sent by Atlassian Jira
(v8.20.10#820010)