Hi Michael,
I've upgraded to stunnel 4.25 and found a few small bugs in the init script.
In stunnel.init.in, I'm pretty sure the DAEMON line should be DAEMON=@prefix@/bin/stunnel and not DAEMON=@prefix@/etc/stunnel
Also, when the parent process is killed with stop, I see Sep 17 14:34:16 server stunnel: warning: can't get client address: Bad file descriptor because the child processes seem to stick around.
I've modified the init script to kill the main process using the pid file and then using pidof to identify and kill the remaining child processes. Is this expected behavior?
regards, KAM
Kevin A. McGrail wrote:
In stunnel.init.in, I'm pretty sure the DAEMON line should be DAEMON=@prefix@/bin/stunnel and not DAEMON=@prefix@/etc/stunnel
Fixed. Thank you.
Also, when the parent process is killed with stop, I see Sep 17 14:34:16 server stunnel: warning: can't get client address: Bad file descriptor because the child processes seem to stick around.
I've modified the init script to kill the main process using the pid file and then using pidof to identify and kill the remaining child processes. Is this expected behavior?
What about: ftp://stunnel.mirt.net/stunnel/libwrap_shutdown.patch Does it help?
Best regards, Mike
Also, when the parent process is killed with stop, I see Sep 17 14:34:16 server stunnel: warning: can't get client address: Bad file descriptor because the child processes seem to stick around.
I've modified the init script to kill the main process using the pid file and then using pidof to identify and kill the remaining child processes. Is this expected behavior?
What about: ftp://stunnel.mirt.net/stunnel/libwrap_shutdown.patch Does it help?
On 4.25, it patched with offset of 1 line:
patching file src/libwrap.c Hunk #1 succeeded at 81 with fuzz 1 (offset -1 lines).
But otherwise the patch does seem to have the desired effect. Looks good to me. I'm using the stock init script with this patch and the daemon change previously mentioned.
Regards, KAM
-
Kevin A. McGrail -
Michal Trojnara