Hello! I can see that there's huge unevenness with regards to striped pool usage: around 80% of load end up on just two striped threads, sys-stripe-11-#12%ITLXCLKAP-server425b5ae8-214a-4ff8-b1a4-943a1513c415% and sys-stripe-9-#10%ITLXCLKAP-server425b5ae8-214a-4ff8-b1a4-943a1513c415%. That's 2/3 of all utilization of all theads.
I imagine there is a lot of contention around a very narrow set of keys. Maybe you update a few keys continuously which will cause locking issues. Please check your data partitioning and affinity. Regards, -- Ilya Kasnacheev чт, 5 сент. 2019 г. в 18:12, Кузин Никита (Nikita Kuzin) < nku...@iitdgroup.ru>: > jfr from ignite is in attachment > > _________________________________ > С уважением, Никита Кузин > Ведущий программист-разработчик > > * Интернейшнл АйТи Дистрибьюшн* > > e-mail: nku...@iitdgroup.ru > тел.: 84995021375 доб. 320 > моб. тел.: 79260948887 > 115114, Москва, Дербеневская ул., 20-27 > ------------------------------ > *От:* Ilya Kasnacheev <ilya.kasnach...@gmail.com> > *Отправлено:* 2 сентября 2019 г. 14:08 > *Кому:* user@ignite.apache.org <user@ignite.apache.org> > *Тема:* Re: Perfomance drop > > Hello! > > If GC is fine next step is examining thread dumps and logs when a long > operation is taking place. You can trigger it yourself by calling > IgniteUtils.dumpThreads() in a test setting. > > Regards, > -- > Ilya Kasnacheev > > > пн, 2 сент. 2019 г. в 11:11, Кузин Никита (Nikita Kuzin) < > nku...@iitdgroup.ru>: > > Hello! > > gc log for 3 days of working ignite. > > There are 58 GCs with duration 100-200ms and 9 GCs with duration of > 200-300ms > > _________________________________ > С уважением, Никита Кузин > Ведущий программист-разработчик > > * Интернейшнл АйТи Дистрибьюшн* > > e-mail: nku...@iitdgroup.ru > тел.: 84995021375 доб. 320 > моб. тел.: 79260948887 > 115114, Москва, Дербеневская ул., 20-27 > ------------------------------ > *От:* Кузин Никита (Nikita Kuzin) <nku...@iitdgroup.ru> > *Отправлено:* 31 августа 2019 г. 10:26 > *Кому:* user@ignite.apache.org <user@ignite.apache.org> > *Тема:* RE: Perfomance drop > > Hello! > > We collected GC log, it seems to be ok. > > Long duration operations are very random events, we will try to collect. > > Can the caches' eviction processes has influence on cache reading? (use of > same threads) > _________________________________ > С уважением, Никита Кузин > Ведущий программист-разработчик > > * Интернейшнл АйТи Дистрибьюшн* > > e-mail: nku...@iitdgroup.ru > тел.: 84995021375 доб. 320 > моб. тел.: 79260948887 > 115114, Москва, Дербеневская ул., 20-27 > ------------------------------ > *От:* Ilya Kasnacheev <ilya.kasnach...@gmail.com> > *Отправлено:* 30 августа 2019 г. 15:49 > *Кому:* user@ignite.apache.org <user@ignite.apache.org> > *Тема:* Re: Perfomance drop > > Hello! > > Have you tried collecting thread dumps/JFR/etc while such an operation is > waiting? It could provide some clues. > > Do you have GC log collected? Any long GC pauses? > > Regards, > -- > Ilya Kasnacheev > > > пт, 30 авг. 2019 г. в 14:43, Кузин Никита (Nikita Kuzin) < > nku...@iitdgroup.ru>: > > Hi! > > We have application that goes to ignite node (only one node in grid). > Application and ignite are located on one machine. > > And we noticed that sometimes that cache located at ignite is very slow: > we execute method igniteCache.containsKey(key) and it can lasts 900ms, size > of cache less than 10000 entries > This cache is used mostly for read. > > Persistence is off > All of JVM recommendations is applied > OS: windows server 2012 R2 > > _________________________________ > С уважением, Никита Кузин > Ведущий программист-разработчик > > * Интернейшнл АйТи Дистрибьюшн* > > e-mail: nku...@iitdgroup.ru > тел.: 84995021375 доб. 320 > моб. тел.: 79260948887 > 115114, Москва, Дербеневская ул., 20-27 > >