Hello all,
I am running Version: 4.1.0-5005 Community Edition (build-5005).
My cluster consist of 5 nodes on centos 7.2 on AWS. The instance type are r3.2xlarge, so 8vcpu and 60 GIGS of RAM.
Replication was configured for 1 bucket to elasticsearch using default options. The bucket size is 64.2GB.
I had several OOM errors in the past because of the goxdcr process.
I started from a 3 nodes with 14 GIG RAM cluster.
I upgraded my cluster to 3 nodes with 30 GIG RAM. I still had the issue.
I am now at 4 nodes running 606 GIG RAM.
I estimate the goxdcr process will be killed in the next hour or so.
Here is the current memory consomption:
top - 18:40:04 up 1 day, 1:12, 6 users, load average: 5.88, 4.21, 3.75
Tasks: 220 total, 3 running, 217 sleeping, 0 stopped, 0 zombie
%Cpu(s): 40.2 us, 5.0 sy, 0.0 ni, 34.8 id, 19.4 wa, 0.0 hi, 0.5 si, 0.2 st
KiB Mem : 62653388 total, 4820196 free, 49820380 used, 8012812 buff/cache
KiB Swap: 10485756 total, 10485756 free, 0 used. 12537832 avail Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1512 couchba+ 20 0 46.852g 0.038t 4768 S 25.2 65.1 80:48.33 goxdcr
1349 couchba+ 20 0 12.944g 5.993g 1896 S 91.4 10.0 718:27.38 mc:listener
1561 couchba+ 20 0 2522644 961724 3976 S 0.7 1.5 10:46.86 indexer
Anyone has an opinion as to what I could do to resolve my issue?
Regards,
Steeve