Problems with the Distributed Persistent Memory Engine

DPME does not start

The DPME service, mvmallocd, fails to start.

Cause: DPME is already running

The DPME service is already running on the system.

Only one mvmallocd process is allowed to run at a time.

Remedy: Check for a running DPME service

Procedure

Check for a running mvmallocd process on the system:

# systemctl status mvmallocd

The system shows a running process:

# systemctl status mvmallocd
● mvmallocd.service - MemVerge Memory Engine
   Loaded: loaded (/etc/systemd/system/mvmallocd.service; disabled; vendor preset: disabled)
   Active: active (running) since Wed 2020-09-23 11:25:41 PDT; 3h 46min ago
  Process: 22672 ExecStart=/usr/bin/bash /usr/libexec/mvmallocd_daemon_start.sh (code=exited, status=0>
  Process: 22650 ExecStartPre=/usr/bin/bash /usr/libexec/mvmallocd_pre_check.sh (code=exited, status=0>
 Main PID: 22684 (mvmallocd)
    Tasks: 2 (limit: 26213)
   Memory: 3.7M
   CGroup: /system.slice/mvmallocd.service
           └─22684 /opt/memverge/sbin/mvmallocd --config=/etc/memverge/mvmallocd.yml

Do one of the following:
- If the running DPME service is working, do nothing.
- If you think there is a problem with the running DPME service, kill the mvmallocd process and start the service again. See Stopping the Distributed Persistent Memory Engine and Starting the Distributed Persistent Memory Engine.

Cause: Domain socket already exists

Attempting to run the DPME service, the following message is logged in the DPME log:

cat /var/log/memverge/mvmallocd_log
-----NEW MVMLOG FILE 2020-12-29 21:05:40
2020-12-29 21:05:40 -0800 [29153:29153]: License loaded successfully.
2020-12-29 21:05:40 -0800 [29153:29153]: Chunk Usage: (3014 / 64508)
2020-12-29 21:05:40 -0800 [29153:29153]: domain socket /tmp/dpme_daemon.0 already exists

In this example, the /tmp/dpme_daemon.0 socket path already exists.

Remedy: Remove the socket and restart the DPME service

Procedure

Verify that the socket path exists:

# ls /tmp/dpme_daemon.0
/tmp/dpme_daemon.0
# rm /tmp/dpme_daemon.0
# rm: remove socket '/tmp/dpme_daemon.0'? y
# ls /tmp/dpme_daemon.0
ls: cannot access '/tmp/dpme_daemon.0': No such file or directory

Remove the socket path:
```
# rm -rf /tmp/dpme_daemon.0
```
Restart the DPME service:
```
# systemctl start mvmallocd
```