debe commented on PR #1155: URL: https://github.com/apache/solr/pull/1155#issuecomment-1344569141
> @debe thank you for taking a look! > > First off, mockito is troublesome. we can take it out completely if there is a way to setup solr without it. I added it only for the transient cores part. same for the lookup misses, it was just an attempt to reproduce the results. > > > I think you actually need to do something while holding the lock like it is in reality. > > I don't completely follow there. what else happens inside the lock block from getCoreFromAnyList? ideally we would test the code as is, I don't see anything else inside that method. by introducing the consumer you are artificially increasing the time spent inside the lock, is this a better mirror of a running system? Ahhh, no I think this is a misunderstanding. I exactly copied the work that is actually done. CoreContainer.getCore() set's incRefCount to true and triggers SolrCore.open(). If you set IncRefCount to false the call to core.open() never happens and therefore everything inside open() which are MDC.put calls (I've counted them) and atomicInt incrementation. > > > I think one should limit the number of threads to the number of CPU cores, otherwise you're benchmarking context switching of OS threads. > > I agree with this point, anything we can tweak on the benchmark settings to get it closer to real life is good for me. > > > To get an overview which public methods from SolrCores and CoreContainer are actually called and how frequent, I've checked a recent flight recorder image from one of our production machines. > > thank you for this data. I would like to use it as a reference for the benchmark, but ideally we would setup the core without any interventions (still not 100% sold on the extra consumer you added). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
