Startup after installing nightly…

Do you overwrite the config file, yes or no?
For I sure dont, only if there are some hefty modifications in the file format.

I dont get the described probem.

@Dave.B wrote:

I just installed 1489 on my Slug – no problems as yet, although it’s only been running for 2 or 3 minutes. But it prompted a question: With every nightly install I’ve done (starting with svn-909) the install has run OK, and then I get prompted to modify the config file then type “opt/etc/init.d/S60mt-daapd” to start the server.
Thing is, every time I do that, it makes a valiant effort at starting up, in that the disk chatters away to itself for about as long as it takes to scan my library, but the server is inaccessible, both to my SoundBridge and PC.
If I then cycle the power on the Slug using the power button, everything then starts up perfectly.

Is this normal behaviour?

Don’t know, but there are some issues with stopping the thing gracefully on the slug. I think tonights nightlies will have a fix for signal handling on the slug, and hopefully it will be able to shutdown and restart gracefully.

You won’t see it the next time you install nightlies, but the time after *that* it should be more graceful. (/me crosses his fingers)

@mas wrote:

Don’t know, but there are some issues with stopping the thing gracefully on the slug. I think tonights nightlies will have a fix for signal handling on the slug, and hopefully it will be able to shutdown and restart gracefully.

You won’t see it the next time you install nightlies, but the time after *that* it should be more graceful. (/me crosses his fingers)

I think you got it fixed. I used to do a killall -9 mt-daapd to get it down. Without the -9 it would create zombies.
Now a normal killall mt-daapd is sufficient. I still do have to send the kill signal to more than one process though. A simple kill will force me to repeat it 3-4x for different processes.

If tonights nightly works right, then killall mt-daapd will send sigterms to all the processes, which should be ignored by all the threads except the process leader, which should initiate a graceful shutdown.

That’s the theory, anyway. We’ll see. Every time I get something working on linuxthreads, it breaks on nptl or bsd threads. Grrr. Signal handling is the suck. It’s almost as unportable as java.

— Ron

@Dave.B wrote:

@rpedde wrote:

there are some issues with stopping the thing gracefully on the slug

I don’t pretend to understand half of the stuff that got posted after my last post, but I just thought I’d add that I upgraded from 1489 to 1512 today and everything worked properly – no need to cycle the power on the slug – it started up first time.

FWIW – 1489 ran continuously for over 52 days (I only shut it down to go on holiday, so I suspect it probably would have gone on indefinitely). My previous record was running 1359 for 22 days before a crash stopped it.

Woohoo… getting there. 🙂 Maybe folks are right and it’s time to start thinking about a new stable…

I had 1463 running for ~80 days until I had a power outage and upgraded to 1498. Stable as hell on my NSLU2, both of them.
😉