Tvheadend

K Shea wrote:

I am running HTS Tvheadend 3.9.2827~g477feab~wheezy and it has always worked great until tonight. I attempted to go into the backend and the moment I went to the Digital Video Recorder tab it crashed. I had to reboot it to get it to come back up. If I have the log open at the bottom of the screen, it says (in red text):

[...]

At around the time it crashes, this appears in /var/log/messages:

Nov 7 22:43:29 satellite kernel: [256481.485413] tcp_server_star¹¹⁶⁴⁴: segfault at 0 ip 00007f4099cc314a sp 00007f40817e1318 error 4 in libc-2.13.so[7f4099bb0000+181000]

This happens EVERY SINGLE TIME I try to go into the DVR tab. All other tabs appear to work as long as I stay away from the DVR tab.

If I am in Kodi I can see the EPG, I can add a timer, and so on. It's only attempting to access the DVR tab in the backend that causes TVHeadEnd to crash. Unfortunately, that's the tab that I use most often, to set up recording schedules.

What weird is nothing has changed on this system recently; in fact the only thing I can think of that has changed is the time, from daylight time back to standard time. But I am fairly certain I have accessed the DVR tab since the time changed.

Is there anything I can do about this short of a complete reinstall?

Why are you using 3.9 when there are the 4.x releases?

Because the OS is Debian Wheezy and there are no newer builds that will work on that.

K Shea wrote:

Because the OS is Debian Wheezy and there are no newer builds that will work on that.

Ah yes, I think your only option is to build the latest from git using the autobuild.sh script.

This gets weirder. I tried everything I could think of last night, including multiple reboots to get this to resolve itself, and it crashed every time I went into that tab. Finally went to bed and now, a few hours later, I find I can go to that tab and it does not crash, and everything looks normal. Truly a WTF moment.

My wild theory #1 is that something in tzdata is somehow triggering this, because it happened on a Saturday night in November and even though the change from daylight to standard time happened last weekend in my part of the world, there may be some localities that don't change until the first weekend in November, and tzdata is always getting updated. I know at first glance this has nothing to do with the libc-2.13.so that seems to have caused the problem, but as far as I can tell, not that much has been updated recently outside of standard Debian/Linux stuff.

My maybe not so wild theory #2 is that it had something to do with the EPG data, which is refreshed every morning between 5 and 6 AM. I have no idea why or how the EPG data could have caused this but that is the only thing that I know of that has changed on this system between when I went to bed and when I got up. Why that would cause an error with libc-2.13.so, I have no idea.

Mark Clarkstone wrote:

K Shea wrote:

Because the OS is Debian Wheezy and there are no newer builds that will work on that.

Ah yes, I think your only option is to build the latest from git using the autobuild.sh script.

Before I would ever attempt that I would do a complete reformat of the hard drive and install Ubuntu Server and a newer version of TVHeadEnd from one of the repositories (probably the "unstable" branch, so I get occasional major updates but not a new version every week). I really hate being stuck in this little slice of upgrade hell because there is no 4.x build of TVHeadEnd for Debian Wheezy, and I'm not knowledgeable enough in Linux to build software from git, plus I'd be afraid of losing all my existing settings and still be stuck at an older version of Debian. Never again will I install Debian for use with TVHeadEnd, since all the upgrade support seems to be for Ubuntu. And I actually do want to upgrade the system eventually; I'd just rather do it at a time when most of the shows are in reruns so that if it takes a couple days to get everything right, I'm not losing recordings.

Also, since libc-2.13.so is a system file, there's no guarantee that upgrading TVHeadEnd from git would even fix the issue (if it ever returns, which I sure hope it doesn't), and it would far more likely that I'd really mess things up if I tried to do that because I'd have no idea what I'm doing.