Hi all,

I have a R5E50 backend running on oldish hardware that's been completely reliable for the past few months. I've made no recent changes to the hardware or software configuration at all (other than to manage recording schedules).

A week ago I found that the backend wasn't running. There was nothing obvious or scary that I could find in the logs, so I just restarted the box. The backend came back after rebooting with no problems, so I just figured it was a fluke.

This morning it happened again. I noticed the problem because I was trying to check on last night's recordings in mythweb, but it told me it couldn't talk to the backend. The machine itself seems to be running fine, I can ssh into it and poke around. But the backend isn't running.

I looked at the logs and am not seeing anything suspicious. Although I do see the following kind of entries repeated once a day at 6:25 AM:


I have no idea if the above log entries are OK, or if they are indicative of a problem. But they're very regular so I assume they're a side effect of a cron task.

The mysql log files are consistently empty. The disk is running at about 83% full:


To get running again, I rebooted my box this morning. Again, the backend came up like a charm. But since this is the second failure in a week, I'm afraid that this will keep happening.

If anyone has any suggestions as to what I can do next to resolve this problem, I'd appreciate it.

_________________
Best,

Allen Cronce

In looking further at the logs, there may be a hint here. There's a weekly event in the daemon.log's like the following:


When I look at the messages around that time, I see the following:


Note the second restart just after the mythconverg log entry. That's two within 20 some odd minutes of each other. I'm not sure what causes the first one.

It was around 7:35 when I noticed that the backend was not running and restarted the system. Of course, that's when I checked the box, so maybe the backend had not been running for some time. Although it did a successful recording 5 hours earlier.

I looked a week ago and see a similar log pattern. There's the "mythconverg checked and backedup" log entry, then sometime after that I notice that mythbackend isn't running.

So maybe something to do with this mythconverg operation is messing up the backend?

Although I have to admit that I'm not sure what I'm seeing here. It isn't clear what the "restart" events are. The certainly aren't a restart of the system, because according to uptime there was no system reboot.

Also, I see similar log entries two weeks ago where there was no backend hickup.

Anyway, my hope is that someone who actually understands these log entries will be able to use them to help figure out this problem.

Thanks in advance for any suggestions.

_________________
Best,

Allen Cronce

Hi,

Until there is a solution, one of the fine folks on the forum has a helper babysitter to keep the backend awake

Mike

Yes, this backend problem (backend won't start or quits randomly on it's own) is a known problem with version R5E50.

No one from the development or testing group has suggested a fix the this bug. I interpret their silence to mean they don't know what the problem is or how to fix it.

I am also running the backend babysitter script as a workaround.
see http://mysettopbox.tv/phpBB2/viewtopic.php?t=13613

Stay sane!

_________________
KnoppMyth R5.5, Asus A8N-VM CSM (nvidia 6150 onboard video), AMD Athlon 64 dual-core 4200+, two 1GB sticks DDR 400, HD-3000 HDTV card, PVR-150 card, Iguanaworks RS-232 IR receiver/transmitter, Pioneer DVR-110 DVD burner

Thanks for the responses. I was aware of the babysitter script. It was suggested when I posted regarding a similar problem where the backend would not start sometimes after a reboot. But this new problem where the backend quits seemed different to me, and I couldn't find people talking about it, so I started this thread.

While I understand that pragmatically the babysitter script is probably the best work around, it bothers me that this problem is not better understood. I don't like the idea of applying bandaids.

And I'm concerned that the babysitter may get in my way someday. If in the future I really want to stop the backend, I now have to remember that it will automatically be restarted by the babysitter.

Anyway, I guess I'll hold my nose and install the babysitter. Hopefully the dev team has enough information to address the real problem in a future release.

Best,

_________________
Best,

Allen Cronce

If I remember correctly there was a version that checked to see if the frontend or myth-setup was running before starting the backend up, those are the only reasons that I have to keep the backend down.

There's a check to make sure that backkup/restore/setup aren't running, and the frontend MUST be running, in order for the babysitter to do anything. Otherwise, it aborts. I did this as it would typically cover most cases where you don't want the backend restarting.

So if you want to do something with the backend down, just exit from the frontend (exit to the desktop) and you can safely work with the backend.

And yes, you're correct, the main problem I was targeting was the backend not coming up after restart. But I also had situations where the backend would die for no apparent reason, about once/month. I had been rebooting nightly. The babysitter seems to be handling that with less disturbance.
--
Chris

Thanks Chris, I was about to post some of your code to show this - but straight from the source is better

Yes, thanks for responding Chris. And thanks for the script. My backend server is headless, but I can live with the frontend running. It should be pretty easy to remember to quit the frontend first.

One question about installation though. From your thread I followed these instructions while logged in as root:

cd /root
crontab -l > crontab.txt
pico crontab.txt

Yes, I'm a pico wimp; vi scares me. But the issue is that the crontab.txt file is empty. I would have thought that there would be tons of cron jobs already in place, but apparently I don't have any. Or rather I don't have any root cron jobs.

Is this expected for a stock R5E50 installation? If so, then I guess I'll add the babysitting script as the only root cron job.

_________________
Best,

Allen Cronce

Nevermind! I just read a little further in the thread and saw the exchange between you and neutron68 regarding root vs. system cron jobs. Looks like not having any root cron jobs is normal, so I'm going to proceed as per the original instructions.

Sorry for not reading ahead!

_________________
Best,

Allen Cronce

You can easily modify the code to check for something else, such as the presence of a file (\root\cancelbabysitter.txt) or the presence or absence of some other job.
Or, just rip out the crontab and put it back later.

It is? I think it is safe to say, I've been running E longer than anyone and have not seen issues with the backend dying on it's own. Perhaps information from the backend log instead of syslog would be more helpful... I've see MARK occur in syslog without detriment.

It is.

TJC wrote about the backend not starting in his "R5E50 how-to"

I tried the mythtv-backend script fix and it didn't help on my system. The backend still dropped randomly.

Thornsoft posted another workaround for the problem. You can see he isn't the only person with a system where the backend stops on it's own.
http://mysettopbox.tv/phpBB2/viewtopic.php?t=13613

I have loaded and reloaded R5E50 onto a second computer at home with the exact same motherboard and processor of my main KnoppMyth box and about 90% of the time, the backend fails to start on it without any alterations following the install.

I will be glad to provide any information that will help you track down the cause of this. I'll see if my logs have anything of use in them. You think this might be hardware specific?

_________________
KnoppMyth R5.5, Asus A8N-VM CSM (nvidia 6150 onboard video), AMD Athlon 64 dual-core 4200+, two 1GB sticks DDR 400, HD-3000 HDTV card, PVR-150 card, Iguanaworks RS-232 IR receiver/transmitter, Pioneer DVR-110 DVD burner

That is a different problem than not starting. As far as I can tell the not starting thing is just a race condition at boot time, it seems to occur around upgrades, possibly because the DB schema is being upgraded. Once my backend is up it stays up without any problems.

Oh, I see. I thought they were one in the same problem.

OK, so what could be causing Thronsoft's (and other people's) system to drop the backend? Cecil said the backend log might have clues. Is there any particular keyword we should look for in the backend log?

_________________
KnoppMyth R5.5, Asus A8N-VM CSM (nvidia 6150 onboard video), AMD Athlon 64 dual-core 4200+, two 1GB sticks DDR 400, HD-3000 HDTV card, PVR-150 card, Iguanaworks RS-232 IR receiver/transmitter, Pioneer DVR-110 DVD burner