A large number of files generated under $GPCC_HOME/ccagent directory in Greenplum

Products

• VMware Tanzu Greenplum

Issue/Introduction

If standard_conforming_strings is set to 'OFF' (Default is ON), the GPCC ccagent module is unable read the .csv log file correctly, and will then generate a dat file under $GPCC_HOME/ccagent at each host.

It will generate 1 file every 1 minute.

In an example case, 13 million files were generated at each host under ccagent directory.
inode use was 100%.

We can use 'df -i' to check the inode usage.

Sample filename:

-rw------- 1 gpadmin gpadmin 0 Sep 22 20:39 ./gpcc_exec_9_2019-09-22_204026.dat
-rw------- 1 gpadmin gpadmin 0 Sep 22 20:44 ./gpcc_exec_9_2019-09-22_204456.dat
-rw------- 1 gpadmin gpadmin 0 Sep 22 21:55 ./gpcc_exec_9_2019-09-22_215556.dat
-rw------- 1 gpadmin gpadmin 0 Sep 22 22:58 ./gpcc_exec_9_2019-09-22_225926.dat

Note: This affects GPCC v4.6 and earlier releases.

Environment

Product Version: 4.1

Resolution

The workaround is to set "standard_conforming_strings" parameter to ON.

The resolution is to upgrade GPCC to v4.7 or later.