@Braiam Oooh! I want to provide the side effect w/r/t MythTV!

I have a HDHomeRun (network ATSC tuner). Just recently signed up for their schedule service, and found the first problem—it appears duplicate checking now works.

I want to record each program several times, because its over an antenna—sometimes there is interference and I need to try a 2nd recording.

So there are options to check for duplicates in current recordings, previous recordings, or current and previous recordings. The docs seem to say that current recordings (obviously enough) means things that currently are sitting on your disks, previous recordings mean ones you've deleted.

And of course, current and previous would be both.

But when I pick previous, it turns out that includes current recordings too! So why is there an option for both? (insert sarcastic remark)

Ok, there is another option to turn off duplicate checking entirely. Ok. Hit save. It does nothing. WTF. Go to a different screen, pull up the same recording edit screen, change it, hit save. Now it takes. (another sarcastic remark)

Fair enough, now I'll just have to mark all old recordings I've already watched as "never record" so I don't get them again if they rerun. Oh wait, now "never record" doesn't work either! You can click it, it pretends its going to work, but the schedule shows it'll record anyway. Apparently you have to have duplicate detection turned on... (probably time to upgrade to sardonic remarks)

And then, this morning the machine found a bad block on one of the RAID disks. So it had an I/O hiccup for 10s or so. What did mythtv do? Here's its log message:

Jul 17 06:32:35 Watt mythlogserver[32650]: mythbackend[32633]: E HDHRStreamHandler threadedfilewriter.cpp:239 (Write) TFW(/srv/mythtv/video/1261_20150717095900.mpg:69): Maximum buffer size exceeded.
                                                                   file will be truncated, no further writing will be done.
                                                                   This generally indicates your disk performance
                                                                   is insufficient to deal with the number of on-going

So I checked; the web frontend still said it was recording.

The light on the tuner was still lit. The switch ports were still flashing with plenty of data.

So it was still recording. Out of paranoia, I checked the file on disk... it didn't truncate it, but it wasn't growing any more. And it was a fair bit smaller than the web interface said.

So it appears MythTV decided to use a far too small buffer (seriously, 100MB would have survived a minute of hiccup), keep using that tuner, lie to itself that it was a successful recording, ... which of course means it wouldn't have re-recorded the program when it re-aired a few hours later. (insert obscene remark)

<hr>

obviously you need to record that in devnull-as-a-service.com

@Braiam LOL, I think MythTV already did!

So... anyone know of a simple program that records from HDHomeRun tuners based on Schedules Direct data? Just recording is fine. I don't need all the fancy UI stuff MythTV does.

@derobert boo

The bot beat you to it.

Hi @derobert. The (insert sarcastic remark) reminds me a little of Buffy. Which of course you have never heard of.

interesting question at very least unix.stackexchange.com/q/216784/41104

@FaheemMitha the one played by Geller?

@derobert Yes, that one.

Well, in that case, I've heard of her. Even watched it sometimes, back during its initial run.

@derobert To be clear, reminds me of Buffy the show, not Buffy the person.

Though of course, the latter was in the former.

I think I've heard of MythTV, but of course, it might be a myth. Like Buffy.

So, @derobert, how's it going? Aside from the TV thing.

@FaheemMitha Oh, other than that, fairly well.

@derobert That's good.

I set up a Debian package for our local logcheck rules, and wrote some scripts to automate building it for way too many versions of Debian. And then added a bunch of rules. Greatly reducing the email load, and finding a few real problems too...

funny

@derobert Why did it need a Debian package?

Lots of machines?

@FaheemMitha Yep. Lots of machines. Makes updating it on all of them easier.

@derobert no puppet?

@derobert ok

or chef, or whatever autodeploy is in the hots right now?

@Braiam Doesn't puppet require packages?

@Braiam Unfortunately not. There are lots of machines, with almost nothing in common between them...

@FaheemMitha not sure

@Braiam ok

And getting one of those running on things as old as Sarge... doesn't seem promising.

(Though, I haven't built my package for Sarge yet. Only Etch and later. Etch!)

anyone ever run gitlab? I was trying it out on a small vps instance (low memory, single core) and most of the time it served 502 errors while the background ruby process kept dying and respawning.

probably due to the restricted memory, but probably not worth running if that was the case

> 512MB RAM + 1.5GB of swap is the absolute minimum but we strongly advise against this amount of memory. See the unicorn worker section below for more advise.

source doc.gitlab.com/ce/install/requirements.html#memory

the instance had 512 mb, so I can agree that I don't advise it either :)

it was overkill for what I wanted anyway

pure plain git server?

a bit more than that. I already have gitweb running one local box that I sync my repos to

@casey I run gitlab here

Ideally I'd like something small scale that has an issue tracker (that can sync with github for my public projects)

anthony-ldap@git:~$ free -m
             total       used       free     shared    buffers     cached
Mem:          2009       1926         83          0          0        655
-/+ buffers/cache:       1271        738
Swap:         2045         89       1956

... so I wouldn't try that on a 512MB VM!

really what I need to do is just setup commit hooks on my local server to push to github automatically

@derobert yea, i didn't really think it would work :)

right now my workflow is split locally and im pushing to two remotes (one the server on my LAN, the other github) and sometimes they aren't in sync

and my free private repos on github will not be free soon, so I want to take them off github but still have my issue trackers :)

I do that at home—the two remotes thing. But often I find that a good thing that they're not in sync. I can push stuff that isn't really ready to publish to my local server (just to have a backup). And when it's ready, push to github/gitlab.

gitlab has a hosted service with free private repositories

@derobert that works usually until I pull from the supercomputer and don't realize I'm not running quite what I want

@derobert yea, I might just do that

and the other day that cost me a few hours where my unoptimized, bad loop order parallelization-does-nothing version was running

amazing what changing loop indices and moving the openmp directives can do

@casey that seems like fun

yea, I don't what I was thinking exactly when I setup the loop originally. The problem is embarrassingly parallel (e.g. benchmarked at 5.1 speedup on my 6 core chip with 6 threads) but due to my errors it was limited to serial performance :)

@casey have you tried kallithea? it's a bit beta, and has mercurial origins, but might work for git. Dunno.

and hyperthreading w/ 12 threads only increased that to 5.3. Its that kind of problem I don't like hyperthreading because if the scheduler doesn't put all 6 threads (in a 6 thread run) onto separate cores, performance takes a big hit

@FaheemMitha i'll take a look

@casey well, you should tell your future self with a comment, next time

:)

kallithea-scm.org

Hello!

I have a question about tags

I just posted this question, but found it very hard to decide which tags to use

There's shell command-line shell-script etc

@FaheemMitha It very often is. What were they on about?

@jacwah That is more of a shell-script question since it is about scripting rather than the general use of a shell. I agree that the distinction could be clearer though.

@terdon What about command-line?

@jacwah That one is for basically anything that has anything to do with the command line. You could tag a question about file manipulation (moving, renaming, copying etc) from the command line with it for example.

It's not that useful here since almost everything we have is about the command line. Not everything, but a huge proportion.

@terdon I forget. It was fairly surreal.

How are you doing? Had a nice vacation?

@FaheemMitha Sun-burnt but otherwise fine thanks :)

And yeah, the ELU room prides itself on being surreal.

Anyway, I just got back today, had a long day and I'm off to bed. Night all.