The dead bodies of the zombies are piling up.
I’m doing inter-node spawns from a program. When
the parent (spawning) process dies, the child continues
to run, of course. If I kill it, it disappears from “ps -A”,
but still shows in “System Information” as a child of “io-net”,
status “unavailable”. There are still entries in /proc for all of
these zombie processes. I don’t seem to be able to flush them
out of the system, short of restarting io-net. Is this a resource
leak in “io-net”?
What do I have to do, restart io-net every once in a while?
John Nagle
Team Overbot
Mike Schneider wrote:
How about adding SPAWN_NOZOMBIE to the inherit flags in the spawn() call?
ms…
John Nagle wrote:
The dead bodies of the zombies are piling up.
I’m doing inter-node spawns from a program. When
the parent (spawning) process dies, the child continues
to run, of course. If I kill it, it disappears from “ps -A”,
but still shows in “System Information” as a child of “io-net”,
status “unavailable”. There are still entries in /proc for all of
these zombie processes. I don’t seem to be able to flush them
out of the system, short of restarting io-net. Is this a resource
leak in “io-net”?
What do I have to do, restart io-net every once in a while?
John Nagle
Team Overbot
Ah, another undocumented l33t feature. Where is it
documented? The help page for “spawn” does not list this flag.
It’s in the QNX 4 migration guide, listed as a deprecated
feature with no explaination. What does it do? Should I use
it?
There still seems to be a bug that io-net doesn’t reap
its children for zombies.
John Nagle
Team Overbot
Mike Schneider wrote:
Mike Schneider wrote:
How about adding SPAWN_NOZOMBIE to the inherit flags in the spawn() call?
ms…
John Nagle wrote:
The dead bodies of the zombies are piling up.
I’m doing inter-node spawns from a program. When
the parent (spawning) process dies, the child continues
to run, of course. If I kill it, it disappears from “ps -A”,
but still shows in “System Information” as a child of “io-net”,
status “unavailable”. There are still entries in /proc for all of
these zombie processes. I don’t seem to be able to flush them
out of the system, short of restarting io-net. Is this a resource
leak in “io-net”?
What do I have to do, restart io-net every once in a while?
John Nagle
Team Overbot
\
I found it in spawn.h.
It worked for me, zombies without it, no zombies with it.
ms…
John Nagle wrote:
Ah, another undocumented l33t feature. Where is it
documented? The help page for “spawn” does not list this flag.
It’s in the QNX 4 migration guide, listed as a deprecated
feature with no explaination. What does it do? Should I use
it?
There still seems to be a bug that io-net doesn’t reap
its children for zombies.
John Nagle
Team Overbot
Mike Schneider wrote:
Mike Schneider wrote:
How about adding SPAWN_NOZOMBIE to the inherit flags in the spawn() call?
ms…
John Nagle wrote:
The dead bodies of the zombies are piling up.
I’m doing inter-node spawns from a program. When
the parent (spawning) process dies, the child continues
to run, of course. If I kill it, it disappears from “ps -A”,
but still shows in “System Information” as a child of “io-net”,
status “unavailable”. There are still entries in /proc for all of
these zombie processes. I don’t seem to be able to flush them
out of the system, short of restarting io-net. Is this a resource
leak in “io-net”?
What do I have to do, restart io-net every once in a while?
John Nagle
Team Overbot
\