Thread: [jgroups-dev] JGroups memory leak?
Brought to you by:
belaban
Menu
▾
▴
javagroups-development
|
From: Development i. <jav...@li...> - 2017-08-29 08:20:09
|
Hi, For the last 2 years, we are using JGroups (currently 3.4.8) for messaging between java processes, some are java main and some running under Tomcat. We are now facing memory issues and used a profiler to analyze them, and found out a huge increase in the # of JGroups classes. See attached stack straces and image We are using the attached configuration. i'll appreciate if someone can tell me what's wrong... Thanks in advance. jgroups_classes.png <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroups_classes.png> stacktraces.txt <http://jgroups.1086181.n5.nabble.com/file/n11406/stacktraces.txt> jgroupsConf.xml <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroupsConf.xml> -- View this message in context: http://jgroups.1086181.n5.nabble.com/JGroups-memory-leak-tp11406.html Sent from the JGroups - Dev mailing list archive at Nabble.com. |
|
From: Development i. <jav...@li...> - 2017-08-29 16:02:07
|
I don't see anything wrong with this: what you're seeing in the screen shot are TcpConnections, and 1 thread for each waiting on I/O... These connections don't use up a lot of memory. It would be more interesting to see "hotspots by object size" and/or "hostspots by object count"... On 29/08/17 10:20, Development issues wrote: > Hi, > For the last 2 years, we are using JGroups (currently 3.4.8) for messaging > between java processes, some are java main and some running under Tomcat. > We are now facing memory issues and used a profiler to analyze them, and > found out a huge increase in the # of JGroups classes. See attached stack > straces and image > > We are using the attached configuration. > > i'll appreciate if someone can tell me what's wrong... > > Thanks in advance. > jgroups_classes.png > <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroups_classes.png> > stacktraces.txt > <http://jgroups.1086181.n5.nabble.com/file/n11406/stacktraces.txt> > jgroupsConf.xml > <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroupsConf.xml> > > > > > -- > View this message in context: http://jgroups.1086181.n5.nabble.com/JGroups-memory-leak-tp11406.html > Sent from the JGroups - Dev mailing list archive at Nabble.com. > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Javagroups-development mailing list > -- Bela Ban | http://www.jgroups.org |
|
From: Development i. <jav...@li...> - 2017-08-30 11:25:28
|
Thanks for your reply. I do not see anything abnormal in "hotspots by object size" or "hostspots by object count", except for some object clones of org.jgroups.protocols.MERGE3$MergeHeader$Type. I was wondering whether all TcpConnections may point on an infinite call loop (of send-receive) that causes multiple classes to be created and eventually cause a leak? On Tue, Aug 29, 2017 at 7:01 PM, Development issues < jav...@li...> wrote: > I don't see anything wrong with this: what you're seeing in the screen > shot are TcpConnections, and 1 thread for each waiting on I/O... > > These connections don't use up a lot of memory. > > It would be more interesting to see "hotspots by object size" and/or > "hostspots by object count"... > > > > On 29/08/17 10:20, Development issues wrote: > >> Hi, >> For the last 2 years, we are using JGroups (currently 3.4.8) for messaging >> between java processes, some are java main and some running under Tomcat. >> We are now facing memory issues and used a profiler to analyze them, and >> found out a huge increase in the # of JGroups classes. See attached stack >> straces and image >> >> We are using the attached configuration. >> >> i'll appreciate if someone can tell me what's wrong... >> >> Thanks in advance. >> jgroups_classes.png >> <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroups_classes.png> >> stacktraces.txt >> <http://jgroups.1086181.n5.nabble.com/file/n11406/stacktraces.txt> >> jgroupsConf.xml >> <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroupsConf.xml> >> >> >> >> >> -- >> View this message in context: http://jgroups.1086181.n5.nabb >> le.com/JGroups-memory-leak-tp11406.html >> Sent from the JGroups - Dev mailing list archive at Nabble.com. >> >> ------------------------------------------------------------ >> ------------------ >> Check out the vibrant tech community on one of the world's most >> engaging tech sites, Slashdot.org! http://sdm.link/slashdot >> _______________________________________________ >> Javagroups-development mailing list >> >> > -- > Bela Ban | http://www.jgroups.org > > > > ------------------------------------------------------------ > ------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Javagroups-development mailing list > > |
|
From: Development i. <jav...@li...> - 2017-08-31 08:00:33
|
On 30/08/17 13:25, Development issues wrote: > Thanks for your reply. > I do not see anything abnormal in "hotspots by object size" or > "hostspots by object count", except for some object clones of > org.jgroups.protocols.MERGE3$MergeHeader$Type. OK, so then what makes you think JGroups is leaking memory? > I was wondering whether all TcpConnections may point on an infinite call > loop (of send-receive) that causes multiple classes to be created and > eventually cause a leak? No, there should be 1 TcpConnection object *per destnation*, e.g. if you have a cluster of 10, and everyone sends messages to everybody else (or multicasts messages), then every member should have 9 TcpConnection objects (each having a reader thread). > On Tue, Aug 29, 2017 at 7:01 PM, Development issues > <jav...@li... > <mailto:jav...@li...>> wrote: > > I don't see anything wrong with this: what you're seeing in the > screen shot are TcpConnections, and 1 thread for each waiting on I/O... > > These connections don't use up a lot of memory. > > It would be more interesting to see "hotspots by object size" and/or > "hostspots by object count"... > > > > On 29/08/17 10:20, Development issues wrote: > > Hi, > For the last 2 years, we are using JGroups (currently 3.4.8) for > messaging > between java processes, some are java main and some running > under Tomcat. > We are now facing memory issues and used a profiler to analyze > them, and > found out a huge increase in the # of JGroups classes. See > attached stack > straces and image > > We are using the attached configuration. > > i'll appreciate if someone can tell me what's wrong... > > Thanks in advance. > jgroups_classes.png > <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroups_classes.png > <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroups_classes.png>> > stacktraces.txt > <http://jgroups.1086181.n5.nabble.com/file/n11406/stacktraces.txt <http://jgroups.1086181.n5.nabble.com/file/n11406/stacktraces.txt>> > jgroupsConf.xml > <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroupsConf.xml <http://jgroups.1086181.n5.nabble.com/file/n11406/jgroupsConf.xml>> > > > > > -- > View this message in context: > http://jgroups.1086181.n5.nabble.com/JGroups-memory-leak-tp11406.html > <http://jgroups.1086181.n5.nabble.com/JGroups-memory-leak-tp11406.html> > Sent from the JGroups - Dev mailing list archive at Nabble.com. > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Javagroups-development mailing list > > > -- > Bela Ban | http://www.jgroups.org > > > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > Javagroups-development mailing list > > > > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > > > > _______________________________________________ > Javagroups-development mailing list > -- Bela Ban | http://www.jgroups.org |
Thanks for helping keep SourceForge clean.
X