I have a simple black diamond workflow that succeeded – according to jobstate.log, pegasus-analyzer and pegasus-status. However, I have a single workflow notification on_error that got triggered nevertheless:

<!-- part 1.1: invocations -->
<invoke when="on_error">/bin/mailx -s 'blackdiamond failed' voeckler</invoke>

$ cat diamond-0.notify
WORKFLOW 261a096a-2ffb-4769-a7b2-36dee49cee2b on_error /bin/mailx -s 'blackdiamond failed' voeckler

$ fgrep -e '***' jobstate.log
1319069879 INTERNAL *** MONITORD_STARTED ***
1319069879 INTERNAL *** DAGMAN_STARTED 20.0 ***
1319070322 INTERNAL *** DAGMAN_FINISHED 0 ***
1319070324 INTERNAL *** MONITORD_FINISHED 0 ***

So, what is going on here?