: We have a pretty standard out-of-the-box solr/jetty setup. Using the web : UI at /solr/admin/logging, for WARNING or SEVERE we get less logging, but : none of CONFIG, FINE or FINEST result in any *more* logging than just at : INFO.
The UI (and the underlying Solr API) lets you change the log levels used to filter what messages are *produced* by the "loggers" - but there is a seperate setting that controls which messages are *accepted* by each "log handler" (ie: you might have multiple log handlers: one writing errors to console, one writting errors and warnings to a "problems.txt" file, and one writing all log messages to a "full_log.txt" file) There is no Solr API/UI for changing the levels on the handlers, it can only be done via the underlying log framework configuration. My understanding is that folks who like using the run time UI for toggling the log levels in Solr typically setup their underlying framework config so that at least one handler is configured to recieve accept at the "FINEST" level, but the root loger is configured to only produce at the "INFO" (or "WARNING") level. Then, when the situation arrises, using the UI to temporarily switch a logger to the FINE or FINEST levels will be written out by that handler. : Is there another place to look for something that might be controlling : that? Maybe a threshold=INFO somewhere? We've been unable to find : anything. FYI: there is a "threshold" param on the /admin/logging handler in 4.0, but (AFAIK) it controls the threshhold of messages recorded by Solr's internal "Log Watcher" for display in the UI screens (not on the console or in log files) : We're trying to turn up logging because our solr indexing server is hanging : at least once a day and we're trying to find out why. It becomes : unresponsive and we have to kill -9 it. it sounds like logging isn't what you should be looking at -- it sounds like you should be trying to generate some thread dumps when the server is not being responsive. -Hoss
