I seemed to have two of my four Zope worker processes (
client3) running an old version -- they seem to have survived my last update (where I simply lifted psycopg2 from 2.5.5 to 2.8.6); no
.pid files for them were present, just
.lock files. (They sometimes complained about not being able to close a closed connection, which should be fixed by psycopg2 v2.7.5).
First thing I did -- I disabled these zombie workers in the (Apache) balancer-manager.
It was not possible to
stop them, so I killed them (no
-9 or the like) and, as soon as
ps -p ... didn't find the process anymore, deleted the .lock files.
starting the two workers again, I found that only one of the looks like expected (with both
.lock file, and the other one (
client3) has a
.lock file only again;
bin/client3 status still tells me, "daemon manager not running". It seems to work, though, telling from accessing its port via
What should I do about this?
Ignore the missing
.pid file and re-enable that worker?
What is common treatment for such abandoned