The monitoring box kept insisting another copy was already running.
Every scheduled MRTG run produced the same ugly result. A few Perl warnings appeared first. Then the useful line: another copy of MRTG seemed to be running, so the new process quit.
The alerts kept arriving. I spent time looking for the process that was starting it, but could not catch anything obvious. I suspected cPanel because other people had reported similar behavior. Most of the forum answers seemed to amount to living with it.
That was not much of an answer when the system kept sending mail.
I wrote back from a plane and promised to have it handled by noon the next day. I would look again after landing, then come back to the larger firewall work later in the week.
The fix was much less interesting than the investigation. A stale lock file made the monitor believe another copy was active. I deleted the stale file and ran it again.
“Should be fixed now,” was the whole useful update.
The update I sent was deliberately small because that is all I can confirm right now. The stale state is gone and the command is running again. The next scheduled run will show whether the duplicate-process message returns.
The Perl warnings pointed at the surrounding configuration, but the line stopping the job was the duplicate-process check. Removing the lock addresses that specific stop condition.
For now: delete the lock, run the monitor, and watch the next result.
That is the check.