EPD Service not starting on ESXi

Products

• VMware vSAN

Issue/Introduction

- This article provides troubleshooting steps to get the EPD service running if it will not start
- EPD (Entry Persistence Daemon) is a user space daemon that runs on every host that is part of the vSAN cluster. The main job of EPD is to make sure there is no component leakage when objects are deleted.

Symptoms:
- EPD service is not running
- Unable to start EPD service 

The epd.log may show the following errors

2024-06-28T20:25:44.324Z No(13) epd[2106465]: EntryDB_EnableCksumExtension: Failed to run VACUUM : database disk image is malformed (11) failedAt:1919
2024-06-28T20:25:44.324Z No(13) epd[2106465]: EntryDB_Open: Failed to enable libcksumvfs db '/scratch/epd-storeV2.db': database disk image is malformed (195887105)
2024-06-28T20:25:44.324Z No(13) epd[2106465]: EPDStoreOpen: Failed to open db (/scratch/epd-storeV2.db): Failure

Environment

ESXi 8.0

VMware vSAN 8.x

Cause

- Scratch partition issues cause the EPD service to fail. The EPD service writes to an epd.db file on /scratch.

If /scratch is not available the EPD service cannot start. 

Resolution

We can copy the current database file to a new one and delete the old one. Then restart the epd.

(1) Stop epd service
/etc/init.d/epd stop

(2) Verify epd is stopped
 /etc/init.d/epd status

(3) Copy existing epd database
cp /scratch/epd-storeV2.db /scratch/epd-storeV2-copy.db

(4) Delete existing epd database
rm -f /scratch/epd-storeV2.db

(5) Restart the epd service and verify the status
/etc/init.d/epd start

/etc/init.d/epd status

Additional Information

Storing ESXi coredump and scratch partitions in vSAN
Unable to start EPD service on vSAN hosts leads to vSAN health object warnings