[
https://issues.apache.org/jira/browse/CONNECTORS-1778?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18023100#comment-18023100
]
Tim Allison commented on CONNECTORS-1778:
-----------------------------------------
Over on TIKA-4494, I showed that that ERROR msg is being caught by POI and is
not causing a restart.
That said, the client for tika-server should be flexible enough to allow for
occasional down time when the child process is restarting. The child process
restarts when "maxFiles" is hit or on OOM or when something else catastrophic
happens.
Let me know if I can help.
> Error: Repeated service interruptions - failure processing document: The
> target server failed to respond
> --------------------------------------------------------------------------------------------------------
>
> Key: CONNECTORS-1778
> URL: https://issues.apache.org/jira/browse/CONNECTORS-1778
> Project: ManifoldCF
> Issue Type: Bug
> Components: Tika extractor
> Affects Versions: ManifoldCF 2.28
> Reporter: mbiso
> Assignee: Piergiorgio Lucidi
> Priority: Major
>
> Hi.
> I have a job ingesting a windows network share.
> It use tika server (standalone)
> There are many errors on Tika because some files cause error like:
>
> {code:java}
> ERROR [qtp131037934-61] 10:44:03,903
> org.apache.poi.xssf.eventusermodel.XSSFSheetXMLHandler Failed to parse SST
> index '1356 '
> java.lang.NumberFormatException: For input string: "1356 " {code}
> The errors cause a restart of a child tika process, and this is reported like
> an interruption in the ManifoldCF job.
> It ends with the message: "Error: Repeated service interruptions - failure
> processing document: The target server failed to respond"
>
> How could I get over this issue?
> I have opened an issue [TIKA-4494 ] on Tika as well, but It could be a right
> behaviour on Tika: many errors cause a restart child process, so this is a
> problem for me.
>
> Any suggestion?
> Thanks a lot.
> Mario Bisonti
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
