[
https://issues.jboss.org/browse/ELY-1323?page=com.atlassian.jira.plugin.s...
]
Jan Kalina commented on ELY-1323:
---------------------------------
There is intentionally audit-endpoints forced to wait for flush on every log event accept
by keeping default *audit-endpoint.synchronized=true*
[~jcacek] would be possible to retest with *synchronized=false*?
Performance issue in audit endpoints
------------------------------------
Key: ELY-1323
URL:
https://issues.jboss.org/browse/ELY-1323
Project: WildFly Elytron
Issue Type: Bug
Components: Audit
Affects Versions: 1.1.0.CR4
Reporter: Josef Cacek
Assignee: Jan Kalina
Priority: Blocker
Attachments: elytron-flightrecording.jfr
Synchronization in audit endpoint implementations leads to a great performance drop.
We see the issue in {{FileAuditEndpoint.accept(EventPriority, String)}}, but similar code
is also in the {{SyslogAuditEndpoint}}.
Check the attached recording (FlightRecorder) in jmc (Java Mission Control).
It can be seen in these tabs
- Threads > Contention
- Threads > Thread Dumps
- Threads > Lock Instances
--
This message was sent by Atlassian JIRA
(v7.2.3#72005)