screen keeps dying

[Turgut Kalfaoglu]

Hi folks. I recently put something like this in my /jfss/scripts/services-start :
/opt/sbin/screen -dmS hep-sync-ns bash -c '/opt/bin/hep-sync-ns.rexx; exec bash'

(I have a name server running on the router, and the hep-sync-ns is a script that sleeps for a while and then syncs with its source DNS server)

But I often find that that screen is "dead" like this:
# screen -r
There is a screen on:
4079.pts-0.ns2 (Dead ???)
Remove dead screens with 'screen -wipe'.
There is no screen to be resumed.

turgut@ns2:/tmp/mnt/2TBEXTERNAL/entware/etc/init.d# screen -wipe
There is a screen on:
4079.pts-0.ns2 (Removed)
1 socket wiped out.
No Sockets found in /opt/tmp/screens/S-turgut.

Why does it die? I have ample swap space defined and enabled even.. The script that is launched and runs in an infinite loop is very simple.
Many thanks, -t

 

[eibgrad]

I guess the natural question to ask is if it's screen itself that's dying, or your script? One thing's for sure; it's not obvious what that script is doing, esp. the exec bash. The answer might lie in having a better understanding of that script. It just doesn't look right. Normally I would place everything in a single script and call that w/ screen. That way you can verify the script works correctly and reliably *before* committing it to screen.

P.S. I assume you installed bash (which isn't a normal part of the Merlin firmware) from Entware into /opt/...? And it's on the path (since you don't specify a path to it in the script)?