Details
-
Bug
-
Resolution: Incomplete
-
Critical
-
4.5.0
-
None
-
Version: 4.5.0-1308 Enterprise Edition (build-1308).
CentOS 7, 64-bit.
-
Untriaged
-
Unknown
Description
A test two node cluster has gotten itself into a state where the indexer and goxdcr process are both runaway on CPU on one node. There is no workload on the system and there hasn't been any for about 12 hours or so.
The node is centos7-1. Note that it's only one of two nodes that are in this state.
The systems have been used under pretty moderate workload.
I've also noted a bunch of core files, surprisingly. Those seem to be unrelated.
Just to give a sense of it, here's top with the threads expanded:
1232 couchba+ 20 0 748012 32504 9132 R 99.9 0.5 4878:10 indexer
|
18726 couchba+ 20 0 209516 5360 3948 R 99.7 0.1 4874:50 goxdcr
|
1241 couchba+ 20 0 748012 32504 9132 R 7.0 0.5 352:33.39 indexer
|
28857 couchba+ 20 0 1949328 476108 3696 S 2.3 8.0 37:06.64 beam.smp
|
28858 couchba+ 20 0 1949328 476108 3696 S 2.0 8.0 37:19.13 beam.smp
|
28855 couchba+ 20 0 1949328 476108 3696 S 1.7 8.0 38:50.61 beam.smp
|
28856 couchba+ 20 0 1949328 476108 3696 S 1.0 8.0 38:22.08 beam.smp
|
18728 couchba+ 20 0 209516 5360 3948 S 0.7 0.1 5:54.17 goxdcr
|
1356 couchba+ 20 0 566228 160732 7576 S 0.3 2.7 81:05.31 mc:nonio_9
|
28843 couchba+ 20 0 1949328 476108 3696 S 0.3 8.0 0:30.18 beam.smp
|
28886 couchba+ 20 0 987980 52852 6264 S 0.3 0.9 3:27.77 beam.smp
|
28897 couchba+ 20 0 10308 6164 1020 S 0.3 0.1 0:30.45 godu
|
790 couchba+ 20 0 11572 568 388 S 0.0 0.0 0:04.96 epmd
|
1065 couchba+ 20 0 1262100 40832 2696 S 0.0 0.7 0:00.01 beam.smp
|