Closed AshenShalitha closed 2 years ago
Hmm, again this is mandatory implementation of org.osgi.service.log.LogReaderService
... Events are collected inside ... synchronized LinkedList, so I can imagine it's not very optimal (even if access to actual appender is also synchronized - depending on the appender of course).
I think you give me no choice (thanks! :)) but to introduce yet another flag...
By the way, I'm going to simply set null
as the service instead of no-op interface implementation - just to save few CPU cycles ;)
@grgrzybek Can we get this fix port into 1.11.x version and get a release.
Hi @tharindu1st, for 1.11.x it's fixed in #493 and released already in 1.11.16 ;)
Hi team,
Recently we encountered a performance issue associated with pax-logging 1.11.3 and raised through [1]. This issue was resolved by introducing a new system parameter
org.ops4j.pax.logging.eventAdminEnabled
. However, in our latest test runs with pax-logging 1.11.15, we could observe that the performance bottleneck is not completely resolved yet and moved to a subsequent line. Please refer to the following stack trace. Is there any configuration to avoid this as well? Please advise.[1] https://github.com/ops4j/org.ops4j.pax.logging/issues/480