I have a MBE / FE and a FE only - both running R5A15.1 with the 7174 Nvidia drivers (installed with the install-nvidia-7174-debian.sh script). The MBE is a P4 2.4Ghz, 512MB RAM, 2 PVR-250's, 64MB GeForce4 MX440. The MB for this machine has a VIA chipset (not sure which version). Intermittently, although seemingly at least once a week, one of the tuners will stop recording mid program. When this happens, mysql AND GDM stop responding. The other tuner will continue to record. A search of /var/log/syslog shows several lines like these:

Oct 27 21:09:12 basemyth kernel: NVRM: Xid: 16, Head 00000000 Count 0006b19a
Oct 27 21:09:14 basemyth kernel: NVRM: Xid: 8, Channel 0000001e
Oct 27 21:09:20 basemyth kernel: NVRM: Xid: 16, Head 00000000 Count 0006b19b
Oct 27 21:09:22 basemyth kernel: NVRM: Xid: 8, Channel 0000001e

/home/mythtv/.xsession-errors shows several lines about not being able to connect to the database (mysql has stopped responding)

/var/log/mythtv/mythbackend.log shows nothing unusual.

None of the log files show anything about any DMA issues. Should I expect to see DMA issues in the logs if this were the VIA / DMA issue? The research I've done into the NVRM errors seem to point to a problem with the drivers...

If the problem does seem to be related to the Nvidia drivers, is it safe to 'downgrade' the drivers using the "install-nvidia-6629-debian.sh" script?

Any help / suggestions / ideas will be appreciated. (I am hesitant to upgrade to R5A22 because the WAF is very high right now, and my setup has gone through months of tweaking - I simply don't have a lot of time to dedicate to setting the system up again),

...forgot to mention that doing the following gets things going again, but I usually end up re-booting.

/etc/init.d/mythtv-backend stop
/etc/init.d/mysql stop
/etc/init.d/mysql start
/etc/init.d/mythtv-backend start
/etc/init.d/gdm restart

If I do manually stop & start he affected services, it will continue to record the show that it was recording when it stopped responding.

Thanks again for any & all help -

Allan Doute

the sympton of the via dma issues is a complete lockup so I don't think that's what's happening to you.

It should be fine to use the 6629 script to "downgrade" nvidia drivers. Are you using XvMC?

_________________
Have a question search the forum and have a look at the KnoppMythWiki.

Xsecrets

Definitely not a complete lock-up. I can SSH in and stop & start all the necessary services. I am not running xvmc. One thing I have noticed is that it seems to happen more often when the box is doing nothing but recording (the frontend is just sitting at the main menu). Every time it happens, tho, the only common log entries are the NVRM ones.

I'll try 'downgrading' the nvidia drivers this weekend - time permitting...

Thanks for your input.

Allan

Well, family life finally gave me enough time (when my MBE wasn't actaully recording anything, and no one was watching anything on the other FE) to actually try the 6629 drivers this morning.

Since I started this post, I've had the problem occur twice. Hopefully, in the next week or so I'll know if "install-nvidia-6629-debian.sh" helped or not.

I'll post again after it's had enough time to prove wether or not it helped.

Thanks again for the input-

Allan

Well, it's been a little over a week (uptime says up 7 days, 9:51) since I 'downgraded' my nvidia drivers from 7174 to 6629, and the backend hasn't missed a single recording.

The backend process did die once - thankfully when nothing was scheduled, and I happened to be browsing Mythweb when it happened - I simply ssh'd in from work & restarted it - everything has been fine ever since. There was nothing unusual in the log files at the time that the mythbackend process died, which kind of seems strange. No NVRAM errors have appeared in any log files since the driver change.

Hopefully the driver change has 'cured' this problem.

Allan

Adoute, I had log messages similiar to yours, although in my case they would also happen with the 6629 driver unless I also used the nvidia AGP driver instead of agpgart. I recently upgraded to nvidia driver version 8756 and after 5.5 days of uptime I have not had the problem, even when using agpgart. Just in FYI in case you feel that upgrading drivers would be useful.


Marc

I had R5B7 (using the default 7174 drivers) running 20 days, then the same errors showed up in the syslog. I tried installing the 6629 drivers (after making sure KM wouldn't try to re-install the 7174 drivers at the next reboot), but they wouldn't install right - so I downloaded and installed the 8178 drivers. So far, I've got an uptime of 12 days.

If I continue to have the issue with these drivers, I'll try switching from agpgart to the nvidia AGP driver. I read your post regarding that change http://mysettopbox.tv/phpBB2/viewtopic.php?t=9027, and will use that for reference if I do need to make the change.

Thanks for the follow up.

Allan

_________________
Running Knoppmyth R5F27 since 09/12/07

Hardware Specs

Good luck. One more thing I did that you may want to be aware of.

1. I've set things up so that the "power" switch on my remote control will do an "/etc/init.d/gdm stop" if mythfrontend is running; it does an "/etc/init.d/gdm restart" if mythfrontend is not running.

2. I have a cronjob that does an "/etc/init.d/gdm stop" at 1:15AM every morning.

This means that many of the front-end process are not running for typically more than half the day -- the same process that appear to be "participating" in the creation of the problem. Just an FYI...

Marc