Endpoint Standard: Repmgr.exe is Performing too Many Disk I/Os

Products

• Carbon Black Cloud Endpoint Standard (formerly Cb Defense)

Issue/Introduction

High Disk utilization or slow performance noted on the developer machines.

Environment

• Endpoint Standard Sensor: All Versions

Cause

• The elevated disk I/O is due to the db rep prune job which continuously checks for file existence for all the files present on disk.
• The frequency of this job also seems quite high running every half hour. In every cycle of this job, it would iterate over all files and would be noticed as continuous disk I/Os in resource monitor.

Resolution

1. Verify if repcli authentication is enabled on the device.
2. If not, enable repcli authentication:  Repcli Authentication on existing sensor 
3. Enable bypass mode in the sensor using the repcli command: How to Toggle Sensor Bypass with the RepCLI Utility
4. Navigate to cfg.ini file and open with Notepad
5. Add "PruneDeletedFilesSleepInterval=43200000"
6. Save changes to cfg.ini
7. Launch elevated command prompt.
8. Run the following repcli command:

"c:\program files\confer\repcli" updateconfig
"c:\program files\confer\repcli" bypass 0  (disable bypass)

Additional Information

• Engineering team is working on a fix related to this issue.
• Since the frequency of pruning job is high, sensor is utilizing the disk just for checking for file existence. As a workaround, reduce the frequency of pruning job so that it would run just twice a day.
• Purning job fequency could be changed as mentioned below:

PruneDeletedFilesSleepInterval=21600000 (6 hours) - This will run 4 times a day.
PruneDeletedFilesSleepInterval=43200000 (12 hours) - This will run twice a day