[ 
https://issues.apache.org/jira/browse/TIKA-5064?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18146366#comment-18146366
 ] 

Hudson commented on TIKA-5064:
------------------------------

UNSTABLE: Integrated in Jenkins build Tika ยป tika-main-jdk17 #1799 (See 
[https://ci-builds.apache.org/job/Tika/job/tika-main-jdk17/1799/])
TIKA-5064 Close HTTP clients when a fetcher is replaced. (#3604) (github: 
[https://github.com/apache/tika/commit/21150843a57f99573797ead0fd8661207a80ae9e])
* (edit) 
tika-pipes/tika-pipes-plugins/tika-pipes-http/src/main/java/org/apache/tika/pipes/fetcher/http/HttpFetcher.java
* (edit) 
tika-pipes/tika-pipes-plugins/tika-pipes-http/src/test/java/org/apache/tika/pipes/fetcher/http/HttpFetcherTest.java
* (edit) 
tika-pipes/tika-pipes-plugins/tika-pipes-atlassian-jwt/src/main/java/org/apache/tika/pipes/fetcher/atlassianjwt/AtlassianJwtFetcher.java


> When SaveFetcher is called regularly, resources leak
> ----------------------------------------------------
>
>                 Key: TIKA-5064
>                 URL: https://issues.apache.org/jira/browse/TIKA-5064
>             Project: Tika
>          Issue Type: Improvement
>            Reporter: Bartek Ciszkowski
>            Priority: Major
>
> HttpFetcher and AtlassianJwtFetcher both maintain HttpClients, but they never 
> close them. When SaveFetcher replaces or DeleteFetcher removes a fetcher, the 
> old instance is never closed, resulting in each client keeping a "Connector 
> evictor" thread alive
> This was discovered when a system calling tika would periodically call 
> SaveFetcher as a means of keeping fetchers in sync and preventing them from 
> going stale. Over time, the resources built up and Tika would OOM.



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

Reply via email to