Hello All, I have installed ser-0.8.11pre29-1 on RedHat 9.0. When I start ser by script /etc/init.d/ser or directly using /usr/sbin/ser, the process SOMETIMES hangs at the Aliases line show below. When I list the process table, I observed there was a defunct process. Now, this does not happen ALL the time when I starts ser. Sometime, /use/sbin/ser or init script will start fine without a hiccup. I also notice that the defunct process is ALWAYS the second lowest process id among all ser processes. Can this be a case of race condition for the parent process to exit without waiting for the child? I did not encounter this using ser-0.8.10 on RedHat 7.3. However, I do remember encountering this behavior once when running ser-0.8.10 on RedHat 9.0, I have removed 0.8.10 and installed 0.8.11 since then for testing 0.8.11. Could this be a problem of RedHat 9.0? Is anyone experience this or am I the only one? [root@linux root]# /usr/sbin/ser Listening on 192.168.1.100 [192.168.1.100]:5060 Aliases: some.where.org:5060 linux.some.where.org:5060 2071 pts/0 S 0:00 /usr/sbin/ser 2072 ? Z 0:00 [ser <defunct>] 2073 ? S 0:00 /usr/sbin/ser 2075 ? S 0:00 /usr/sbin/ser 2076 ? S 0:00 /usr/sbin/ser 2077 ? S 0:00 /usr/sbin/ser 2078 ? S 0:00 /usr/sbin/ser 2079 ? S 0:00 /usr/sbin/ser Thanks for the great software guys. Tsang Han Wong
On Jun 06, 2003 at 22:04, Wong Tsang Han <tsanghan@gmx.net> wrote:
Hello All,
I have installed ser-0.8.11pre29-1 on RedHat 9.0. When I start ser by script /etc/init.d/ser or directly using /usr/sbin/ser, the process SOMETIMES hangs at the Aliases line show below. When I list the process table, I observed there was a defunct process. Now, this does not happen ALL the time when I starts ser. Sometime, /use/sbin/ser or init script will start fine without a hiccup. I also notice that the defunct process is ALWAYS the second lowest process id among all ser processes. Can this be a case of race condition for the parent process to exit without waiting for the child?
I did not encounter this using ser-0.8.10 on RedHat 7.3. However, I do remember encountering this behavior once when running ser-0.8.10 on RedHat 9.0, I have removed 0.8.10 and installed 0.8.11 since then for testing 0.8.11. Could this be a problem of RedHat 9.0? Is anyone experience this or am I the only one?
I'm afraid you are the only one :-) We haven't encoutered this before. Could you send us the config file and could you also look in the syslog for lines starting with ser? (it might log some messages to syslog when it dies). Thanks, Andrei
On 06-06 16:53, Andrei Pelinescu-Onciul wrote:
On Jun 06, 2003 at 22:04, Wong Tsang Han <tsanghan@gmx.net> wrote:
Hello All,
I have installed ser-0.8.11pre29-1 on RedHat 9.0. When I start ser by script /etc/init.d/ser or directly using /usr/sbin/ser, the process SOMETIMES hangs at the Aliases line show below. When I list the process table, I observed there was a defunct process. Now, this does not happen ALL the time when I starts ser. Sometime, /use/sbin/ser or init script will start fine without a hiccup. I also notice that the defunct process is ALWAYS the second lowest process id among all ser processes. Can this be a case of race condition for the parent process to exit without waiting for the child?
I did not encounter this using ser-0.8.10 on RedHat 7.3. However, I do remember encountering this behavior once when running ser-0.8.10 on RedHat 9.0, I have removed 0.8.10 and installed 0.8.11 since then for testing 0.8.11. Could this be a problem of RedHat 9.0? Is anyone experience this or am I the only one?
I'm afraid you are the only one :-) We haven't encoutered this before.
Could you send us the config file and could you also look in the syslog for lines starting with ser? (it might log some messages to syslog when it dies).
I have seen this problem already. It seems to happen on RH 9 only. I have it on my todo list and plan to investigate it later. My guess is that there is some problem with glibc libraries in RH 9 because this happens also with older versions of ser and on RH 9 only. There are no error messages in syslog when this happens. Jan.
On Saturday 07 June 2003 00:31, Jan Janak wrote:
I have seen this problem already. It seems to happen on RH 9 only. I have it on my todo list and plan to investigate it later. My guess is that there is some problem with glibc libraries in RH 9 because this happens also with older versions of ser and on RH 9 only.
There are no error messages in syslog when this happens.
Can this be related to the fact that RedHat build their 9.0 against the Native POSIX Thread Library? Greetings Nils
On Jun 07, 2003 at 00:43, Nils Ohlmeier <nils@iptel.org> wrote:
On Saturday 07 June 2003 00:31, Jan Janak wrote:
I have seen this problem already. It seems to happen on RH 9 only. I have it on my todo list and plan to investigate it later. My guess is that there is some problem with glibc libraries in RH 9 because this happens also with older versions of ser and on RH 9 only.
There are no error messages in syslog when this happens.
Can this be related to the fact that RedHat build their 9.0 against the Native POSIX Thread Library?
No, we don't use threads. Andrei
On Jun 06, 2003 at 16:53, Andrei Pelinescu-Onciul <pelinescu-onciul@fokus.fraunhofer.de> wrote:
On Jun 06, 2003 at 22:04, Wong Tsang Han <tsanghan@gmx.net> wrote:
Hello All,
I have installed ser-0.8.11pre29-1 on RedHat 9.0. When I start ser by script /etc/init.d/ser or directly using /usr/sbin/ser, the process SOMETIMES hangs at the Aliases line show below. When I list the process table, I observed there was a defunct process. Now, this does not happen ALL the time when I starts ser. Sometime, /use/sbin/ser or init script will start fine without a hiccup. I also notice that the defunct process is ALWAYS the second lowest process id among all ser processes. Can this be a case of race condition for the parent process to exit without waiting for the child?
I did not encounter this using ser-0.8.10 on RedHat 7.3. However, I do remember encountering this behavior once when running ser-0.8.10 on RedHat 9.0, I have removed 0.8.10 and installed 0.8.11 since then for testing 0.8.11. Could this be a problem of RedHat 9.0? Is anyone experience this or am I the only one?
I'm afraid you are the only one :-) We haven't encoutered this before.
Well, Daniel manage to reproduce this on RH 9. It seems there is some kind of race condition between our daemonize function and our signal handlers (daemonize will fork & exit to become session leader and we caught our own SIGCHLD). It should be fixed now on cvs (pre31). Could you try again? Thanks, Andrei
participants (4)
-
Andrei Pelinescu-Onciul -
Jan Janak -
Nils Ohlmeier -
Wong Tsang Han