<div dir="ltr">On Tue, Apr 1, 2014 at 11:21 PM, faginbagin <span dir="ltr"><<a href="mailto:mythtv@hbuus.com" target="_blank">mythtv@hbuus.com</a>></span> wrote:<br><div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="">
<br>
</div>I have also been having problems with mythfilldatabase. I found --dd-grab-all worked some of the time, depending on the time of day. Next I looked at my log files, which weren't that helpful. Then I moved /usr/bin/mythfilldatabase to the side and put a shell script in its place that would run the binary mythfilldatabase with debug logging options, when executed by mythbackend. I found that, in my case, the root cause is that downloads from schedules direct are timing out, especially if mythfilldatabase is run shortly after 2am or 9pm (pre DST) or 10pm (post DST). But sometimes, I get a partial download with some <fault...> tags. Since its not a mythtv bug, I reported my problems and observations on the schedules direct support forum. If you want the gory details and want to compare your experience to mine, here's the forum thread:<br>
<a href="http://forums.schedulesdirect.org/viewtopic.php?f=5&t=2541" target="_blank">http://forums.schedulesdirect.org/viewtopic.php?f=5&t=2541</a><br>
<br></blockquote><div><br></div><div>This was what Ian Evans, Phil Bridges and I seemed to agree on in the February thread (Automatic mythfilldatabase...) - download errors occurring around 0200 UTC. To illustrate the problem I've attached a script and output from my backend. The errors really started up in January. My days of guide data was getting into the single digits too often to ignore.<br>
<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Schedules direct seems to think the problem is actually with Tribune's servers, the true source of the listings data. While I wait for a solution from either Tribune or schedules direct, I am experimenting with cron jobs run a different times of day, trying to find the sweet spot of a time of day that works more often than not.<br>
<br></blockquote><div><br></div><div>My theory is that with the old mfdb run window being 2-5 before it was changed to 0-23, a lot of people started hitting that 0200(UTC change bug?) time and TMS blacklisted us for that time.<br>
<br></div><div>Whatever the cause, to mitigate failing runs I filed bug #12101. You can see my results with the patch over the last week in the attached file.<br><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I do think mythfilldatabase could do a better job figuring out what is and is not a successful run. It has been reporting success, even when every download attempt in a given run has timed out and no new listings data has been added to my system.<br>
</blockquote><div><br></div><div>Agreed. Seeing "Success" at the bottom of mythweb or in mythfrontend gives the user a false sense of security.<br> <br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
HTH,<br>
Helen<br></blockquote></div></div></div>