Hello! I can see that it did see one of this messages, but why do you think it is stuck? For what I'm seeing, it has promptly un-stuck in one second: 2020-05-01 12:27:23.386 PDT priority='WARN' thread='Test worker' org.apache.ignite.internal.processors.cache.persistence.IgniteCacheDatabaseSharedManager@127 - Page-based evictions started. Consider increasing 'maxSize' on Data Region configuration: DefaultDataRegion 2020-05-01 12:27:23.387 PDT priority='WARN' thread='Test worker' org.apache.ignite.internal.processors.cache.persistence.evict.Random2LruPageEvictionTracker@127 - Too many attempts to choose data page: 5000 2020-05-01 12:27:24.374 PDT priority='TRACE' thread='Test worker' org.apache.ignite.internal.processors.cache.GridCacheMapEntry@91 - markObsolete0 [key=KeyCacheObjectImpl [part=-1, val=reentrant-select count(t."id") from "sample" t where 1=1 AND t."other"=? AND t."id"=?|bbb291c45a1-5c55-4bf1-86b3-dd8e0c410695, hasValBytes=true], entry=29960847, clear=false]
I would maybe expect a IgniteOOM error after this, but it does not seem to materialize. Why do you think it is stuck for good? However, this message is throttled so it can be spending any amount of time in that code. I can see that you have partition map exchange at the same time, so that's where you might be waiting. I have also found https://issues.apache.org/jira/browse/IGNITE-12510, is there a chance it is relevant? How large are your entries? I also think 512M is a tiny size for data region, try increasing it. Regards, -- Ilya Kasnacheev пт, 1 мая 2020 г. в 23:08, scottmf <scott...@gmail.com>: > out.multipart-aa > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-aa> > > out.multipart-ab > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-ab> > > out.multipart-ac > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-ac> > > out.multipart-ad > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-ad> > > out.multipart-ae > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-ae> > > out.multipart-af > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-af> > > out.multipart-ag > < > http://apache-ignite-users.70518.x6.nabble.com/file/t1632/out.multipart-ag> > > > hi Ilya, I turned on debugging for ignite and dumped the output into a > multipart set of files that i've attached. Let me know if you need anymore > info. If needed I can try to reproduce this in a generic setting but that > will take time. > > since 5MB is the limit, i had to upload the files in 5MB chunks. To > assemble them, put them into a directory then run 'cat * > file.gz' > > Answers to your questions: > > > Do you see any "Too many attempts to choose data page" or "Too many > failed > > attempts to evict page" messages in your logs? > > See output file > > > How large are your data regions > > we only use the default data region with default settings - 512MB > > > how many caches do they have? > > maybe 20ish? > > > I would expect that behavior if eviction can't find any page to evict, if > > all data pages are evicted already and only metadata pages remain, ones > > that cannot be evicted. > > Could you elaborate on this or point me to any docs? > > > > > > > -- > Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >