[ https://issues.apache.org/jira/browse/NUTCH-2890?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sebastian Nagel updated NUTCH-2890: ----------------------------------- Attachment: fetcher-protocol-okhttp-NUTCH-2890-async-profiler.png > Protocok-okhttp: upgrade okhttp to 4.9.1 to address infinite connection > retries > ------------------------------------------------------------------------------- > > Key: NUTCH-2890 > URL: https://issues.apache.org/jira/browse/NUTCH-2890 > Project: Nutch > Issue Type: Bug > Components: plugin, protocol > Affects Versions: 1.18 > Reporter: Sebastian Nagel > Assignee: Sebastian Nagel > Priority: Major > Fix For: 1.19 > > Attachments: fetcher-protocol-okhttp-NUTCH-2890-async-profiler.png > > > In rare situations and under heavy load (> 100 fetcher threads) I've seen > that the protocol-okhttp plugin gets into a state where a significant amount > of CPU time is spent in handling failed connections, see > [async-profiler|https://github.com/jvm-profiling-tools/async-profiler] > screenshot. This is caused by > [okhttp#5819|https://github.com/square/okhttp/issues/5819] which is fixed > with okhttp [4.5.0|https://square.github.io/okhttp/changelog/#version-450]). > We should upgrade to the okhttp 4.9.1 (the current stable version). -- This message was sent by Atlassian Jira (v8.3.4#803005)