[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16679078#comment-16679078 ] Li commented on CASSANDRA-14495: What if the memory usage already hits over 95% and still keeps growing? No latency or throughput impact yet. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16661664#comment-16661664 ] Chris Lohfink commented on CASSANDRA-14495: --- High heap usage is expected and not likely an issue. If it was an issue you would notice other problems (like horrible latencies and many timeouts). > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16661010#comment-16661010 ] Abdul Patel commented on CASSANDRA-14495: - i only see high heap memory usage nothing else, and when i searched for GCInspector i found the the listed message > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16660862#comment-16660862 ] Chris Lohfink commented on CASSANDRA-14495: --- If your gc time is set to 500ms like the default g1 settings are it doesn't mean much just that the JVM is doing what its supposed to do - it fills up enough eden regions and tries to set the number of regions such that with the current allocation rates it will take up the targeted pause time. Take a look at https://www.oracle.com/technetwork/tutorials/tutorials-1876574.html and the gc logs, theres many youtube presentations and blogs that can help walk through the phases and how to read the logs. Do you have an actual problem your experiencing? Bad latencies? timeouts? If so thats different and nodetool tablestats and schema helpful if its a data model issue but try to describe the problem your having and perhaps move this to user list or stack overflow as this jira is for bug reports, new features etc and changes to C* source. Your GCs are fairly frequent though so if its impacting your system people can help identify a bad data model and maybe some mitigation approaches but there are better forums to reach out for that kind of help. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16660666#comment-16660666 ] Abdul Patel commented on CASSANDRA-14495: - I see below GC inspector messages, can some help me undertsnad these ? INFO [Service Thread] 2018-10-22 11:10:22,178 GCInspector.java:284 - G1 Young Generation GC in 251ms. G1 Eden Space: 729808896 -> 0; G1 Old Gen: 7029653504 -> 7138390512; G1 Survivor Space: 109051904 -> 75497472; INFO [Service Thread] 2018-10-22 11:16:05,708 GCInspector.java:284 - G1 Young Generation GC in 209ms. G1 Eden Space: 729808896 -> 0; G1 Old Gen: 4353687552 -> 4483710984; INFO [Service Thread] 2018-10-22 11:16:08,759 GCInspector.java:284 - G1 Young Generation GC in 253ms. G1 Eden Space: 729808896 -> 0; G1 Old Gen: 4483710984 -> 4601151480; INFO [Service Thread] 2018-10-22 11:20:26,201 GCInspector.java:284 - G1 Young Generation GC in 217ms. G1 Eden Space: 729808896 -> 0; G1 Old Gen: 3602907144 -> 2015954944; INFO [Service Thread] 2018-10-22 11:20:34,467 GCInspector.java:284 - G1 Young Generation GC in 206ms. G1 Eden Space: 729808896 -> 0; G1 Old Gen: 1874853896 -> 2052537840; INFO [Service Thread] 2018-10-22 11:48:32,697 GCInspector.java:284 - G1 Young Generation GC in 573ms. G1 Eden Space: 10200547328 -> 0; G1 Old Gen: 35824 -> 0; G1 Survivor Space: 100663296 -> 939524096; INFO [Service Thread] 2018-10-22 11:48:38,038 GCInspector.java:284 - G1 Young Generation GC in 793ms. G1 Eden Space: 3078619136 -> 0; G1 Old Gen: 0 -> 907214328; G1 Survivor Space: 939524096 -> 394264576; INFO [Service Thread] 2018-10-22 11:48:39,133 GCInspector.java:284 - G1 Young Generation GC in 294ms. G1 Eden Space: 461373440 -> 0; G1 Old Gen: 907214328 -> 1291845632; G1 Survivor Space: 394264576 -> 75497472; INFO [Service Thread] 2018-10-22 11:49:12,405 GCInspector.java:284 - G1 Young Generation GC in 222ms. G1 Eden Space: 2181038080 -> 0; G1 Old Gen: 3481272304 -> 3677068784; G1 Survivor Space: 251658240 -> 243269632; INFO [Service Thread] 2018-10-22 11:49:34,485 GCInspector.java:284 - G1 Young Generation GC in 210ms. G1 Eden Space: 4085252096 -> 0; G1 Survivor Space: 67108864 -> 234881024; INFO [Service Thread] 2018-10-22 11:49:41,027 GCInspector.java:284 - G1 Young Generation GC in 208ms. G1 Eden Space: 2290089984 -> 0; G1 Old Gen: 4903141368 -> 5096079352; G1 Survivor Space: 192937984 -> 100663296; INFO [Service Thread] 2018-10-22 11:49:47,059 GCInspector.java:284 - G1 Young Generation GC in 229ms. G1 Eden Space: 2113929216 -> 0; G1 Old Gen: 5096079352 -> 5179965448; G1 Survivor Space: 100663296 -> 260046848; INFO [Service Thread] 2018-10-22 11:49:47,864 GCInspector.java:284 - G1 Young Generation GC in 240ms. G1 Eden Space: 595591168 -> 0; G1 Old Gen: 5179965448 -> 5456591864; G1 Survivor Space: 260046848 -> 41943040; INFO [Service Thread] 2018-10-22 11:51:55,126 GCInspector.java:284 - G1 Young Generation GC in 682ms. G1 Eden Space: 10208935936 -> 0; G1 Old Gen: 2657805472 -> 2663677936; G1 Survivor Space: 92274688 -> 830472192; INFO [Service Thread] 2018-10-22 11:52:02,632 GCInspector.java:284 - G1 Young Generation GC in 614ms. G1 Eden Space: 2558525440 -> 0; G1 Old Gen: 2663677936 -> 3467692024; G1 Survivor Space: 830472192 -> 318767104; INFO [Service Thread] 2018-10-22 11:52:04,595 GCInspector.java:284 - G1 Young Generation GC in 213ms. G1 Eden Space: 536870912 -> 0; G1 Old Gen: 3467692024 -> 3783262192; G1 Survivor Space: 318767104 -> 83886080; INFO [Service Thread] 2018-10-22 11:53:41,556 GCInspector.java:284 - G1 Young Generation GC in 279ms. G1 Eden Space: 10150215680 -> 0; G1 Old Gen: 3793081848 -> 3797276144; G1 Survivor Space: 150994944 -> 662700032; INFO [Service Thread] 2018-10-22 11:53:51,744 GCInspector.java:284 - G1 Young Generation GC in 521ms. G1 Eden Space: 7440695296 -> 0; G1 Old Gen: 3797276144 -> 3918572016; G1 Survivor Space: 662700032 -> 998244352; INFO [Service Thread] 2018-10-22 11:53:52,370 GCInspector.java:284 - G1 Young Generation GC in 589ms. G1 Eden Space: 8388608 -> 0; G1 Old Gen: 3918572016 -> 4907335664; G1 Survivor Space: 998244352 -> 33554432; INFO [Service Thread] 2018-10-22 11:54:31,296 GCInspector.java:284 - G1 Young Generation GC in 253ms. G1 Eden Space: 1937768448 -> 0; G1 Old Gen: 3246391280 -> 3661627384; G1 Survivor Space: 394264576 -> 100663296; INFO [Service Thread] 2018-10-22 11:54:42,711 GCInspector.java:284 - G1 Young Generation GC in 210ms. G1 Eden Space: 872415232 -> 0; G1 Old Gen: 3955228664 -> 4215275512; G1 Survivor Space: 260046848 -> 67108864; INFO [Service Thread] 2018-10-22 11:54:47,487 GCInspector.java:284 - G1 Young Generation GC in 207ms. G1 Eden Space: 3816816640 -> 0; G1 Survivor Space: 67108864 -> 260046848; INFO [Service Thread] 2018-10-22 11:54:48,615 GCInspector.java:284 - G1 Young Generation GC in 212ms. G1 Eden Space: 763363328 -> 0; G1 Old Gen: 4215275512 -> 4466933752; G1 Survivor Space:
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16641911#comment-16641911 ] Chris Lohfink commented on CASSANDRA-14495: --- long GCs or frequent GCs caused by large partitions would cause problems, yes makes sense. The heap usage can be at 5% everytime you look and still have that though in that scenario (as heap fills up, goes down, fills up, goes down depending on when you look). Its not an indicator of a problem in itself, even on a perfectly healthy system with no issues you can check and see the heap usage high. Look at the GC logs or GCInspector reports to see issues. If real curious use the swiss java knive (sjk) and {{java -jar sjk.jar -p CASSANDRA_PID}} and you can see the allocation rate and which threads are doing the allocations. When your pushing 1gb/s objects on heap you will see issues (hardware dependent on where its a problem). > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16641094#comment-16641094 ] Ma Dega commented on CASSANDRA-14495: - Chris, I had a similar problem and Full GC was very frequent and at times there were nodes failures, compaction also never finished. Once I cleared out the large partitions, issue subsided. But the large partitions somehow got produced after the upgrade - if that makes sense. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16641085#comment-16641085 ] Chris Lohfink commented on CASSANDRA-14495: --- {{nodetool garbagecollect}} is unrelated to the jvm GCs. Heap usage is a meaningless metric in itself to track, look at GC times in the logs (search for GCInspector). The heap usage being 70% or 10% doesnt impact anything except when the next GC will be. If your GC time is long or frequent then you have a problem, if your heap usage is high, your heap usage is high and that means nothing. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16640770#comment-16640770 ] Abdul Patel commented on CASSANDRA-14495: - Also is nodetool info best place to cheq heap usage ? > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16640766#comment-16640766 ] Abdul Patel commented on CASSANDRA-14495: - How frequently you had full gc job? Nodedool garbagecollect right? > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16640749#comment-16640749 ] Ma Dega commented on CASSANDRA-14495: - Look for large partitions. Those creep up during compaction "Writing large partition". In my case, once I dealt with these. Constant Full GC and OOM was resolved. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16640544#comment-16640544 ] Chris Lohfink commented on CASSANDRA-14495: --- > heap memory usage bumps up This is how the jvm work, objects created sit on heap and build up until a GC. Heap usage going up is expected normal behavior. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16640541#comment-16640541 ] Abdul Patel commented on CASSANDRA-14495: - i have seen same pattern in 3.11.3 as well , it works for 2-3 weeks and suddenly heap memory usage bumps up. and then every hr i get alerts. the only new thing is , i am installing cassandra reaper as well with new patch, but even with repaer down , i see same behavior. do we just bump of max heap ? or is it bug > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16591123#comment-16591123 ] Ma Dega commented on CASSANDRA-14495: - [~aby786] [~cnlwsu] Looks like I'm experiencing a similar problem with 3.11.2 except in mine I see long GC pauses even in G1GC. Any idea how to optimize this to run through GC quicker? Would adding XX:InitiatingHeapOccupancyPercent=55 help this situation? INFO [Service Thread] 2018-07-29 11:31:35,604 GCInspector.java:284 - G1 Young Generation GC in 517ms. G1 Eden Space: 2306867200 -> 0; G1 Old Gen: 15267266816 -> 15594422528; G1 Survivor Space: 243269632 -> 318767104; INFO [Service Thread] 2018-07-29 11:31:37,408 GCInspector.java:284 - G1 Young Generation GC in 510ms. G1 Eden Space: 1560281088 -> 0; G1 Old Gen: 15594422528 -> 16005464320; G1 Survivor Space: 318767104 -> 234881024; INFO [Service Thread] 2018-07-29 11:31:39,815 GCInspector.java:284 - G1 Young Generation GC in 569ms. G1 Eden Space: 2348810240 -> 0; G1 Old Gen: 16005464320 -> 16278094088; G1 Survivor Space: 234881024 -> 327155712; > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16553017#comment-16553017 ] Abdul Patel commented on CASSANDRA-14495: - no Gc logs and have to reboot the cluster every 2 weeks, anyone else faced the issue ? is it better to wait for 3.11.3? > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16517453#comment-16517453 ] Chris Lohfink commented on CASSANDRA-14495: --- something like {{fgrep "stopped: [1-9]" gclog}} > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16517325#comment-16517325 ] Abdul Patel commented on CASSANDRA-14495: - i have gc log , it doesnt show much of full messages, could you guide on whats best way to check if GC or long pauses are reported? > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16517313#comment-16517313 ] Abdul Patel commented on CASSANDRA-14495: - We still got 80% and abover memory usage alert after a weeks time, whats the next stable version post 3.11.2? > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16517317#comment-16517317 ] Chris Lohfink commented on CASSANDRA-14495: --- As mentioned before, 80% memory usage is not a problem if theres no full GCs, its how java works. Its not a problem. There is nothing wrong and nothing to fix. If you are seeing repeated GCs or long pauses then yes, but the heap filling up is how its supposed to work. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16503646#comment-16503646 ] Abdul Patel commented on CASSANDRA-14495: - I have lowered this to InitiatingHeapOccupancyPercent 30 for now, as default is 45%. I havent seen any memory consumption so far, but its been 3 days now, will monitor for a 8 to 10 days and see if it pops back in. I do agree apart from memory consumption no performance degradation or user compalints have been recieved so faar. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500633#comment-16500633 ] Chris Lohfink commented on CASSANDRA-14495: --- If you dont have long gc pauses or OOM exceptions I wouldn't start trying to tune the GC settings if I were you. Your solving a problem you don't have. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500541#comment-16500541 ] Abdul Patel commented on CASSANDRA-14495: - Its says below for G1GC: ## Optional G1 Settings # Save CPU time on large (>= 16GB) heaps by delaying region scanning # until the heap is 70% full. The default in Hotspot 8u40 is 40%. #-XX:InitiatingHeapOccupancyPercent=70 > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500440#comment-16500440 ] Chris Lohfink commented on CASSANDRA-14495: --- More CPU than necessary possibly in exchange of keeping heap more empty. Honestly I dont think you need to do anything, I personally like that value lower just to reduce impact on fragmentation of the old space when using CMS, if your using G1 it has completely different meaning. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500403#comment-16500403 ] Abdul Patel commented on CASSANDRA-14495: - got it , you are recommeding this parameter to be set @ 55? what would be downside of it? -XX:InitiatingHeapOccupancyPercent=70 > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500398#comment-16500398 ] Chris Lohfink commented on CASSANDRA-14495: --- So what exactly is the problem? Your concerned because in the past the heap usage was lower and now its higher but it actually causes no issues? For what its worth, the way the JVM works - its expected behavior for this to slowly creep up over time. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500389#comment-16500389 ] Abdul Patel commented on CASSANDRA-14495: - I am more concern on why this sudden behaviour in this version only ? could you please share inputs on how i can work on =decrease initiating occupancy (55% I'd recommend) ? we already have G1GC setup, anything else can be done > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500378#comment-16500378 ] Chris Lohfink commented on CASSANDRA-14495: --- Do you have GC thrashing and OOM exceptions or just worried about high utilization? With a 75% (default) initiating occupancy and probably a few gb of YG you would perfectly expect the heap memory to reach over 80 or even 90%. Thats not an issue, but perfectly expected and functioning behavior. If it really concerns you (it shouldn't) You can decrease initiating occupancy (55% I'd recommend) and kick off old gen cleanup earlier. With G1 you can increase the ReserveSpace to like 15% to keep it under 85%. > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500366#comment-16500366 ] Abdul Patel commented on CASSANDRA-14495: - Attached the output , out of 20GB 14Gb is used on one of node for now > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > Attachments: cas_heap.txt > > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Commented] (CASSANDRA-14495) Memory Leak /High Memory usage post 3.11.2 upgrade
[ https://issues.apache.org/jira/browse/CASSANDRA-14495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16500321#comment-16500321 ] Chris Lohfink commented on CASSANDRA-14495: --- Can you include heap histogram (jmap -histo CASSANDRA_PID}}) to see whats in the heap?}} > Memory Leak /High Memory usage post 3.11.2 upgrade > -- > > Key: CASSANDRA-14495 > URL: https://issues.apache.org/jira/browse/CASSANDRA-14495 > Project: Cassandra > Issue Type: Bug > Components: Metrics >Reporter: Abdul Patel >Priority: Major > > Hi All, > > I recently upgraded my non prod cassandra cluster( 4 nodes single DC) from > 3.10 to 3.11.2 version. > No issues reported apart from only nodetool info reporting 80% usage . > I intially had 16GB memory on each node, later i bumped up to 20GB, and > rebooted all nodes. > Waited for an week and now again i have seen memory usage more than 80% , > 16GB + . > this means some memory leaks are happening over the time. > Any one has faced such issue or do we have any workaround ? my 3.11.2 version > upgrade rollout has been halted because of this bug. > === > ID : 65b64f5a-7fe6-4036-94c8-8da9c57718cc > Gossip active : true > Thrift active : true > Native Transport active: true > Load : 985.24 MiB > Generation No : 1526923117 > Uptime (seconds) : 1097684 > Heap Memory (MB) : 16875.64 / 20480.00 > Off Heap Memory (MB) : 20.42 > Data Center : DC7 > Rack : rac1 > Exceptions : 0 > Key Cache : entries 3569, size 421.44 KiB, capacity 100 MiB, > 7931933 hits, 8098632 requests, 0.979 recent hit rate, 14400 save period in > seconds > Row Cache : entries 0, size 0 bytes, capacity 0 bytes, 0 hits, 0 > requests, NaN recent hit rate, 0 save period in seconds > Counter Cache : entries 0, size 0 bytes, capacity 50 MiB, 0 hits, 0 > requests, NaN recent hit rate, 7200 save period in seconds > Chunk Cache : entries 2361, size 147.56 MiB, capacity 3.97 GiB, > 2412803 misses, 72594047 requests, 0.967 recent hit rate, NaN microseconds > miss latency > Percent Repaired : 99.88086234106282% > Token : (invoke with -T/--tokens to see all 256 tokens) -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org