On Thu, Jul 30, 2020 at 02:52:16PM -0700, Thomas Dineen wrote:
Gentle People:
I have a question about "The Curious Mystery Of Killed Process At 4:56
AM"? For some while now I have been developing a optimization
application, which depending on the machine may run for several days.
The development process has played out over a month or more of
testing, finding many bugs including several Segmentation Fault
Crashes, which are usually found and fixed with the help of GDB.
Now I have a new one: After running for several days the
application terminates, and prints "Killed" in the shell window. Now
know this did not likely occur via the keyboard because of physical
security. Can a process kill itself? Is this simply caused by a bug
which must be chased down? Could this be caused by a shortage of
resources? Memory? Threads? Is there a limit on the number of
threads?
I was thinking of user limits. Check the "ulimit" command
for a list of limits imposed on user processes. For example,
my current open files limit is 1024. Are you opening files
and not closing unneeded ones?
A limit new to me is "signals pending". Might you be ignoring
signals to avoid being killed? Maybe something is generating
signals to your process and they queued until the limit was
exceeded.
jon
--
Jon H. LaBadie jonfu(a)jgcomp.com