You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The performance investigation of the tests in source-postgres revealed that java connector containers could require the better part of a second to do something as basic as respond to a spec API call. See #31298 (comment)
Part of what causes the latency to be high is setting up the log machinery. SLF4J version 1.7.x, which is what we currently use, performs an expensive method call that version 2.x skips.
The method in question is findPossibleStaticLoggerBinderPathSet in LoggerFactory and essentially looks up a class in all of the jars on the classpath. Of course part of the problem is that we have a lot of jars ( see #31400 ) but this call is wasteful. It accounts for about 100ms on my laptop.
If all we have to do is bump a few dependencies and tweak a few settings then we should do it.
The text was updated successfully, but these errors were encountered:
Topic
No response
Relevant information
The performance investigation of the tests in source-postgres revealed that java connector containers could require the better part of a second to do something as basic as respond to a
spec
API call. See #31298 (comment)Part of what causes the latency to be high is setting up the log machinery. SLF4J version 1.7.x, which is what we currently use, performs an expensive method call that version 2.x skips.
The method in question is
findPossibleStaticLoggerBinderPathSet
inLoggerFactory
and essentially looks up a class in all of the jars on the classpath. Of course part of the problem is that we have a lot of jars ( see #31400 ) but this call is wasteful. It accounts for about 100ms on my laptop.If all we have to do is bump a few dependencies and tweak a few settings then we should do it.
The text was updated successfully, but these errors were encountered: