[mythtv-users] backend no mythpreviewgen or deletes
jerome at supernet.ab.ca
Thu Oct 27 18:50:30 UTC 2016
On Thursday, October 27, 2016 09:11:22 AM Michael T. Dean wrote:
> On 10/26/2016 01:48 PM, Jerome Yuzyk wrote:
> > [0.27.6, Fedora 22]
> > My backend has been behaving a little oddly over the past couple days and I'm trying to understand why.
> > First off, mythpreviewgen sporadically doesn't work, usually in stretches of recordings, and then starts again. It's curious, a bit inconvenient, but I'd like to know.
> > Secondly, mythbackend stopped slow-deleting deleted recordings a couple of days ago, which I detected with a reporting tool I run nightly. So I restarted the mythbackend service and logged this message:
> > Oct 26 09:21:41 tv mythbackend: 2016-10-26 09:21:41.472032 E AutoExpire: Filesystem
> > Info cache is empty, unable to calculate necessary parameters.
> This happens when the system can't find the directories (and, therefore,
> the file systems) specified by your Storage Groups.
> > I went into the frontend to try to Delete Forever the Deleted recordings that had accumulated and was unsuccessful at first. I'd select Delete Forever and nothing would happen initially for most recordings. So I restarted the mythbackend service again, let it settle, then went back into the frontend to delete and had some success. I kept trying and finally I could Delete Forever each of the Deleted items.
> > The only other oddity is that I had a commflag job that had died with an "Exit Code 136", which got cleared on the second mythbackend restart.
> > Any ideas about what this is all about? The box has been running for only 90 days, which is trivial in its uptime history.
> I have no idea what's actually happening, but it sounds like some
> file-system funkiness--like your file systems are appearing and/or
> disappearing out from under MythTV (and, perhaps, taking too long in
> some cases to reappear, causing a timeout/failure). Are you using some
> kind of auto-mounter for these file systems?
I just mount with fstab, real standard.
I didn't mention that mythfilldatabase didn't run for a couple days either, so there was something systemic going on to hold up these processes.
I Googled the error message and it's been encountered before but there wasn't much to say except that it has happened with multiple mythbackend processes started accidentally. I don't have that problem.
On Thursday, October 27, 2016 02:15:38 PM Stuart Auchterlonie wrote:
> Might also be worth forcing a filesystem check, if this system has
> been around as long as you are inferring.....
> You also didn't mention what filesystem you use here?
I use EXT4. 90 days is not a long uptime though, but yes things can happen. If I see another hint of trouble I'll reboot and see if a FS-check is triggered.
Right now all seems to be OK. The box spent a few hours slow-deleting 200G, I recorded a half-dozen programs that got commflagged and previews generated, and this morning mythfilldatabase ran to bring me up to date. This weekend's sports recordings will be another test.
A little of Jerome's MythTV World: http://mythtv.bss.ab.ca
More information about the mythtv-users