19 Apr
2010
19 Apr
'10
10:44 a.m.
Iñaki Baz Castillo wrote:
I have not yet provide with a solution for this case. If the select
timeout is finished, then the main process will try to write to a pipe
that the other end stop reading, receive EPIPE and kill all its
children. This is the case in witch the main process reaches the write
part, if it's blocked somewhere else... well this must be taken care of.
2010/4/19 marius zbihlei
<marius.zbihlei(a)1and1.ro>ro>:
Hello
I have attached to this email a new main.c.patch containing the select
changes (timeout is 30 seconds, hard coded :) )
I have taken the liberty to also format your initial patch.
Indeed, main process is the process after the
fork, and this is the
process that writes to signal the parent. I see two possible pitfalls:
1. If the main process blocks, this will block
the parent process also
This could be improved using select() with a timeout, right?
2. If the main
process returns without writing the bytes, and there are
still child processes left(tcp or udp worker children etc), then they will still
have the writing part of the socket open (forked from the main process)and again
the parent (master) process will keep blocking (didn't discovered a case where it
might happen).
Then the workers should immediately close the pipe output fd after
being created, right?
Good question.. We can kill all children from the
main proces, but I am not
sure that from the masetr process we can do this..
But if we just kill the main process from the parent process, then all
the workers and rest of children would be killed, right?
The problem would be if the main process gets "blocked" for some
reason, but I think that is a terrible case I've never seen.
Regards.
Cheers
Marius