Closed eroux closed 6 years ago
Current cache settings should solve the issue (at least I wasn't able to reproduce the issue locally during all the profiling sessions I launched, even though I can't really simulate very heavy loads on the system). We can now watch runtime cache and memory info (See http://localhost:8080/cache). The PS Survivor Space memory and PS Old Gen pool tells us about objects not yet garbage collected, so if any memory leak occurs, we should be able to detect them. (Note that PS Old Gen pool contains objects that are garbaged collected less often, due to a longer life cycle).
well, that's really good! I was thinking about something more simple that we could implement but it's not really a priority, just logging, in the java code things like "fetching object XXX in cache", "successuflly retrieved XXX in cache", "pushing XXX in cache", etc. in debug log level. But we can do that later, the new system should have enough features
Actually, I did implement that service thinking we would also need it for IIIF image server (especially to check the way hymir is behaving regarding memory and cache)
I'm still not quite sure why lds-pdi doesn't work, but my current guess is that it has problems with the cache (as the request never reaches fuseki), it would be helpful to have some options to log cache events, as maybe we'll see some timeouts...