-rw-r--r-- 1 gpadmin gpadmin 139G Nov 19 10:39 /usr/local/greenplum-cc-web-4.8.0/logs/agent.log
Files starts as expected:
2019-10-10 13:22:47 [INFO]connect to rpc server ddlgpmtst11a.us.dell.com:8899 2019-10-10 13:22:47 [INFO]ControlCommand: connection estabilshed ddlgpmtst11a.us.dell.com:8899 2019-10-10 13:22:47 [INFO]agent begin to report health status ddlgpstst001.us.dell.com 4d65822107fcfd52 2019-10-10 13:27:47 [INFO]Throughput profile 0msg/s, 0KB/s, #BATCH:1, #FAIL:0, #RETRY:0 2019-10-10 13:32:47 [INFO]Throughput profile 0msg/s, 0KB/s, #BATCH:0, #FAIL:0, #RETRY:0 2019-10-10 13:37:47 [INFO]Throughput profile 0msg/s, 0KB/s, #BATCH:1, #FAIL:0, #RETRY:0
But finishes with streaming error:
2019-11-19 11:14:11 [ERROR]Accept error: accept unixpacket /tmp/.s.GPMC.sock: accept4: too many open files 2019-11-19 11:14:11 [ERROR]Accept error: accept unixpacket /tmp/.s.GPMC.sock: accept4: too many open files 2019-11-19 11:14:11 [ERROR]Accept error: accept unixpacket /tmp/.s.GPMC.sock: accept4: too many open files 2019-11-19 11:14:11 [ERROR]Accept error: accept unixpacket /tmp/.s.GPMC.sock: accept4: too many open files 2019-11-19 11:14:11 [ERROR]Accept error: accept unixpacket /tmp/.s.GPMC.sock: accept4: too many open files 2019-11-19 11:14:11 [ERROR]Accept error: accept unixpacket /tmp/.s.GPMC.sock: accept4: too many open files
The error message contributes to ~99.9% of the file:
wc -l agent.2019-11-19_112826.log 1406762139 agent.2019-11-19_112826.log grep -c "too many open files" agent.2019-11-19_112826.log 1406713968
This issue may be attributed to a low limit on open files.
All servers had 262k set to the following limit:
open files (-n) 262144
We recommend 524288 as per the GPDB documentation.
Here are the recommended limits on open files:
soft nofile 524288
hard nofile 524288
soft nproc 131072
hard nproc 131072