Beirdobot, irc.freenode.net

Saturday, January 25th, 2014, 00:20 UTC

[00:20:14]

MartinT:

anyone want to go halfsies on a hitman to torture and kill the guy who invented css?

[00:23:28]

TheCras5 (TheCras5!~TheCrashe@p508070D7.dip0.t-ipconnect.de) has joined #mythtv

[00:26:35]

TheCrasher (TheCrasher!~TheCrashe@p5DCE4D38.dip0.t-ipconnect.de) has quit (Ping timeout: 252 seconds)

[00:57:41]

gremlyn (gremlyn!~gremlyn@ip72-197-35-46.sd.sd.cox.net) has joined #mythtv

[00:57:57]

gremlyn:

I am, and have been for some time, affected by this seemingly fixed bug: https://github.com/MythTV/packaging/pull/34

[00:58:23]

gremlyn:

yet it does not appear that the has made it into the repo I am updating from...

[00:58:43]

gremlyn:

since it renders mythtv completely useless, I have sadly not been using it, but would really like to

[00:58:50]

gremlyn:

any ideas on what I can do?

[01:45:30]

robink_ (robink_!~quassel@unaffilated/robink) has joined #mythtv

[01:47:30]

robink (robink!~quassel@unaffilated/robink) has quit (Ping timeout: 245 seconds)

[02:15:02]

skd5aner:

gremlyn: is there a ticket in trac associated with the bug?

[02:15:48]

skd5aner:

also, that seems to be a mythbuntu error – might want to ask in #mythbuntu

[02:16:00]

skd5aner:

or,oops – #ubuntu-mythtv

[02:16:33]

skd5aner:

and... it was merged in to master, which isn't released... sooooo...

[02:16:55]

gremlyn:

guess I have to wait for the release...

[02:17:22]

gremlyn:

:\

[02:17:25]

gremlyn:

ok

[02:19:11]

skd5aner:

gremlyn: actually...

[02:19:31]

skd5aner:

gremlyn: it was also commited to 0.27 before release – so, if you try 0.27, it should be fixed there

[02:27:21]

NightMonkey (NightMonkey!~NightrMon@pdpc/supporter/professional/nightmonkey) has quit (Quit: Body blow! Body blow!)

[02:28:26]

gremlyn:

skd5aner, I am on0.27

[02:28:33]

gremlyn:

+fixes IIRC

[02:29:07]

skd5aner:

https://github.com/MythTV/packaging/commits/fixes/0.27 & ndash; look at Sept18

[02:29:28]

gremlyn:

hrm

[02:29:39]

gremlyn:

just confirmed I am definitely on 0.27+fixes

[02:30:26]

gremlyn:

maybe it will magically fix itself when 14.04 drops

[02:32:13]

danielk22 (danielk22!~danielk22@96.57.9.142) has left #mythtv ()

[03:06:02]

MartinT (MartinT!~smuxi@46-18-104-220.static.vivaciti.org) has quit (Remote host closed the connection)

[03:06:07]

gremlyn (gremlyn!~gremlyn@ip72-197-35-46.sd.sd.cox.net) has left #mythtv ()

[03:15:38]

_nyloc_ (_nyloc_!~quassel@p3EE2DE1C.dip0.t-ipconnect.de) has joined #mythtv

[03:19:26]

nyloc (nyloc!~quassel@p3EE2C742.dip0.t-ipconnect.de) has quit (Ping timeout: 264 seconds)

[03:47:07]

peper03 (peper03!~peper03@mythtv/developer/peper03) has quit (Read error: Operation timed out)

[03:52:48]

peper03 (peper03!~peper03@mythtv/developer/peper03) has joined #mythtv

[04:03:53]

TheCras5 (TheCras5!~TheCrashe@p508070D7.dip0.t-ipconnect.de) has quit (Quit: KVIrc 4.3.1 Aria http://www.kvirc.net/)

[04:31:20]

gregg4567 (gregg4567!~kurt@adsl-98-71-166-218.gsp.bellsouth.net) has joined #mythtv

[04:54:59]

fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has quit (Ping timeout: 252 seconds)

[04:56:20]

fetzerch (fetzerch!~quassel@unaffiliated/fetzerch) has joined #mythtv

[05:38:02]

DouglasK (DouglasK!~douglask@135-23-116-76.cpe.pppoe.ca) has joined #mythtv

[05:38:21]

DouglasK:

In Mythweb, I get an error on attempt to load the page... info is at http://pastebin.com/WKKprBkF ... help?

[05:40:13]

DouglasK:

question is, where do I set the mythweb password as opposed to the client password.

[05:43:08]

gregg4567 (gregg4567!~kurt@adsl-98-71-166-218.gsp.bellsouth.net) has left #mythtv ()

[05:58:30]

DouglasK:

In the mythtv home folder, pass looks right, in the /etc/mythtv folder, it's right, in /var/www/mythtv/configuration, it's right too.

[07:10:19]

DouglasK is now known as DouglasKAway

[07:12:00]

knightr (knightr!~Nicolas@mythtv/developer/knightr) has quit (Quit: Leaving)

[07:49:39]

doev (doev!~doev@p4FD433AF.dip0.t-ipconnect.de) has joined #mythtv

[08:21:24]

knightr (knightr!~Nicolas@69-165-170-178.dsl.teksavvy.com) has joined #mythtv

[08:21:25]

knightr (knightr!~Nicolas@69-165-170-178.dsl.teksavvy.com) has quit (Changing host)

[08:21:25]

knightr (knightr!~Nicolas@mythtv/developer/knightr) has joined #mythtv

[08:22:31]

doev (doev!~doev@p4FD433AF.dip0.t-ipconnect.de) has quit (Ping timeout: 260 seconds)

[08:23:33]

joki (joki!~joki@p54861DF5.dip0.t-ipconnect.de) has quit (Read error: Operation timed out)

[08:29:33]

joki (joki!~joki@p54860FB3.dip0.t-ipconnect.de) has joined #mythtv

[09:15:46]

Tobbe5178 (Tobbe5178!~asdf@h104n2-sv-a13.ias.bredband.telia.com) has joined #mythtv

[09:19:58]

doev (doev!~doev@p4FD433AF.dip0.t-ipconnect.de) has joined #mythtv

[09:36:39]

dekarl:

gremlyn skd5aner: it looks like that is a fix in the routine to install a fresh database with 0.27... So to get the fix you have to drop and reinstall the database... (the commits only change deb/debian/mythtv_0.27.0.sql)

[09:36:58]

dekarl:

so there will not be a magic fix that gets applied by waiting and updating

[10:14:37]

amessina (amessina!~amessina@2001:470:c1dc:7779:d6be:d9ff:fe8d:7c1e) has quit (Quit: Konversation terminated!)

[10:17:12]

amessina (amessina!~amessina@2001:470:c1dc:7779:d6be:d9ff:fe8d:7c1e) has joined #mythtv

[10:45:31]

peper03:

stuartm: Ok, I'll take a look at moving them to external.

[11:09:04]

jya (jya!~jyavenard@mythtv/developer/jya) has joined #mythtv

[11:11:39]

jya (jya!~jyavenard@mythtv/developer/jya) has quit (Read error: Connection reset by peer)

[11:12:10]

jya (jya!~jyavenard@mythtv/developer/jya) has joined #mythtv

[11:14:39]

jya (jya!~jyavenard@mythtv/developer/jya) has quit (Read error: Connection reset by peer)

[11:15:36]

jya (jya!~jyavenard@mythtv/developer/jya) has joined #mythtv

[11:27:57]

jya (jya!~jyavenard@mythtv/developer/jya) has quit (Read error: Connection reset by peer)

[11:33:04]

jya (jya!~jyavenard@mythtv/developer/jya) has joined #mythtv

[11:38:15]

jya (jya!~jyavenard@mythtv/developer/jya) has quit (Read error: Connection reset by peer)

[11:43:53]

stuartm:

if only there was a way we could sign the code so that packagers would be thwarted when they tried to link system libs instead of our bundled copies

[12:29:56]

dekarl:

inject a mandatory custom function? Like "IContainMythTVFixesThatAreNotYetUpstream(){return true;}"? Any refer to them in our startup?

[12:33:25]

Steve-Goodey (Steve-Goodey!~steve@host109-158-212-221.range109-158.btcentralplus.com) has joined #mythtv

[12:59:15]

gregL (gregL!~greg@cpe-74-76-105-205.nycap.res.rr.com) has quit (Ping timeout: 260 seconds)

[13:13:48]

stuartm:

they can just patch out such stuff

[13:14:18]

stuartm:

law unto their own :(

[13:16:01]

stuartm:

I just find it highly amusing that they sign all packages so that you can be sure they haven't been modified in any way, yet it's the packagers themselves who are modifying code before it's packaged, and with what oversight, who audits their changes?

[13:17:58]

stuartm:

not good enough to get yourself commit access on a project? No problem! Just become a packager instead and you do whatever you like ... defeats the whole code quality, peer review and security model of open source :(

[13:18:42]

** stuartm is still in a bad mood after an argument with a Mageia packager last night **

[13:24:04]

stichnot (stichnot!~stichnot@adsl-68-127-209-56.dsl.pltn13.pacbell.net) has joined #mythtv

[13:24:04]

stichnot (stichnot!~stichnot@mythtv/developer/stichnot) has joined #mythtv

[13:24:04]

stichnot (stichnot!~stichnot@adsl-68-127-209-56.dsl.pltn13.pacbell.net) has quit (Changing host)

[13:24:31]

clever:

just make --version report that its been modified and doesnt match git perfectly, though that may be harder with a bare source tar

[13:25:17]

gregL (gregL!~greg@cpe-74-76-105-205.nycap.res.rr.com) has joined #mythtv

[13:31:26]

gregL (gregL!~greg@cpe-74-76-105-205.nycap.res.rr.com) has quit (Ping timeout: 264 seconds)

[13:36:46]

dekarl:

stuartm: of course they can patch it out, but they have to shuffle patches whicht remove "dont Touch This Packagers!!!" from the code which should raise some eyebrows in peer review.

[13:37:26]

dekarl:

either way, we are going to take the blame from the users because we tell them "the packagers messed up, take it to them and leave us alone"

[13:37:58]

dekarl:

maybe we can take a peek at how xbmc&others do it.

[13:38:24]

** dekarl thinks about a wall of shame... **

[13:43:06]

dekarl:

Would be nice if they could get smolt working... as of now I'd say its not worth to invest any time into Mageia with not a single registered installation. But a look at their forum hints that there are some installations out there.

[14:09:51]

stuartm:

dekarl: not one? If so I'd say there's something broken there, because I use mageia and so do a few other users

[14:10:13]

stuartm:

smolt is working here

[14:10:17]

Myth_noob (Myth_noob!~richard@87-194-20-182.bethere.co.uk) has joined #mythtv

[14:10:36]

stuartm:

or was ... not sure when I last saw it doing the update

[14:10:38]

Myth_noob (Myth_noob!~richard@87-194-20-182.bethere.co.uk) has left #mythtv ()

[14:11:44]

stuartm:

but Mageia isn't the only one switching bundled libs for system ones, debian et al have all been known to do it

[14:31:00]

TheCrasher (TheCrasher!~TheCrashe@p508070D7.dip0.t-ipconnect.de) has joined #mythtv

[14:39:37]

dekarl:

Debian is known to patch a lot more in software... It appears as if they don't care about security

[14:39:59]

dekarl:

random search hit http://blog.cmpxchg8b.com/2013/08/security-debianisms.html

[14:43:27]

dekarl:

stuartm, can you post the output of "sendProfile.py --printOnly" on Mageia somewhere? And if they enable or disable sending by default.

[14:45:44]

stuartm:

dekarl: http://pastebin.com/iH7efESU

[14:45:56]

stuartm:

which would explain why I've not seen it running recently

[14:52:09]

gregL (gregL!~greg@cpe-74-76-105-205.nycap.res.rr.com) has joined #mythtv

[14:53:31]

brfransen (brfransen!~brfransen@64.179.169.226) has quit (Ping timeout: 260 seconds)

[14:55:40]

brfransen (brfransen!~brfransen@64.179.169.226) has joined #mythtv

[15:09:29]

dekarl:

oops

[16:02:56]

knightr (knightr!~Nicolas@mythtv/developer/knightr) has quit (Read error: Connection reset by peer)

[16:26:15]

NightMonkey (NightMonkey!~NightrMon@173-164-139-193-SFBA.hfc.comcastbusiness.net) has joined #mythtv

[16:26:15]

NightMonkey (NightMonkey!~NightrMon@173-164-139-193-SFBA.hfc.comcastbusiness.net) has quit (Changing host)

[16:26:15]

NightMonkey (NightMonkey!~NightrMon@pdpc/supporter/professional/nightmonkey) has joined #mythtv

[16:35:05]

skd5aner:

stuartm: hey, wasn't sure if you could gleem anything out of the backtrace I posted last night?? Anything you see in it that points to socket issues – or something else? it's the instability I see about 10 times a day now...

[16:36:08]

dekarl1 (dekarl1!~dekarl@p4FE84B64.dip0.t-ipconnect.de) has joined #mythtv

[16:36:40]

dekarl (dekarl!~dekarl@p4FE845ED.dip0.t-ipconnect.de) has quit (Ping timeout: 252 seconds)

[16:36:42]

stuartm:

missed it, will take a look

[16:41:34]

skd5aner:

stuartm: thanks, I'm starting to feel bad that I've tapped you on the shoulder for so long about my lingering issues, but I'm grateful you've been willing to take a couple mins here and there to try to come up with some diagnosis

[16:41:45]

dekarl1 is now known as dekarl

[16:44:50]

skd5aner:

stuartm: fyi, that backtrace was taken on the masterbackend at the time the frontend was experiencing fluttering and unresponsivieness

[16:45:30]

skd5aner:

also, I added a new frontend a few weeks ago – same issues exhibited there.

[16:46:41]

stuartm:

still using a slave backend?

[16:48:49]

skd5aner:

stuartm: well, when the master poops out, the slave never reconnects unless I manually restart mythbackend on the salve...

[16:49:09]

skd5aner:

stuartm: so, it's still "there", but I got really tired of resetting it once or twice a day

[16:49:20]

skd5aner:

right now, the master says it's not connected

[16:50:03]

skd5aner:

probably has been "disconnected" for at least 5 or so days. Occasionally I'll restart it and it'll reconnect. I do think mythbackend is technically running on the slave, just not "connected"

[16:52:07]

stuartm:

gigem: the backtrace skd5aner posted last night seems to be a scheduler deadlock

[16:53:33]

skd5aner:

stuartm: I've had suspicions for a while that I'm having scheduler issues (even back to 0.26) – I frequently see mysqld maxing out a few cores when the scheduler is running, and usually mythbackend as well

[16:53:54]

skd5aner:

stuartm: I've also noticed that I'll get some unresponsiveness on occasion for several seconds after trying to delete recordings

[16:54:23]

skd5aner:

stuartm: not sure if that's because it triggers the scheduler or not... also, I think when there's auto-expires (from deleted recordings or otherwise)

[16:54:27]

stuartm:

skd5aner: have you done a database repair recently?

[16:54:52]

skd5aner:

stuartm: yes, in fact, I think I run the script to check the database daily in cron

[16:55:06]

stuartm:

skd5aner: the scheduler stalling or just taking a very long time might explain some of this

[16:56:17]

stuartm:

you've got half a dozen threads just waiting on the scheduler to finish running, eventually it may even exhaust the thread pool or connection limit after which who knows exactly what happens

[16:56:39]

skd5aner:

stuartm: does the backtrace you just looked at look similiar to the one posted in #11929?

[16:57:28]

skd5aner:

I think that backtrace was taken during similiar exhibited behavior – just curious if it seems similiar or different

[16:57:36]

skd5aner:

#11929

[16:57:36]

** MythLogBot http://code.mythtv.org/trac/ticket/11929 **

[16:57:45]

stuartm:

could even explain why the slave backend is failing, it waits so long for the backend to respond that it times out and then gets stuck in that error condition (because no-one has tested the error handling lately)

[16:58:41]

** skd5aner really wishes that he had the slightest clue on how to read and translate a backtrace **

[16:59:36]

stuartm:

skd5aner: not exactly the same, but again two threads waiting on the scheduler, a third thread is waiting on one of those

[16:59:54]

skd5aner:

ok, so... some similarities at least...

[17:00:12]

skd5aner:

thanks again for checking – I wonder why it's so impactful to me?

[17:00:42]

skd5aner:

I mean, my database is nearly 10 years old... so I have a large number of rules, filters, old recordings, etc...

[17:01:05]

dekarl:

could the sbe<->mbe issue have a similar root cause as #11862? mythmediaserver segfaulting on client disconnects...

[17:01:05]

skd5aner:

I wonder if it's just that the scheduler has to deal with some much that it just isn't "scaling" well

[17:02:47]

skd5aner:

it's gotten progressively worse over the last year or so, and it just seemed like initially of 0.27 there was some minor issues that started to grow, but since then it's gotten progressively worse.

[17:03:28]

skd5aner:

but, I had mentioned to gigem maybe as far back as 0.25 that mythfilldatabase and the scheduler seem to run excessively long (compared to other reports)

[17:04:15]

stuartm:

gigem would have a better idea than I do about what's excessive and also about the locking issues involving the scheduler

[17:05:01]

skd5aner:

yup, makes sense... thanks for helping to hone the compass in the right direction

[17:05:13]

skd5aner:

:)

[17:06:54]

stuartm:

a lockless scheduler would certainly have fewer issues but that would mean interrupting or discarding scheduler results if something changed while it was being run

[17:07:20]

stuartm:

skd5aner: are you using EIT?

[17:07:33]

skd5aner:

no, schedules direct – and I've disabled EIT

[17:07:51]

stuartm:

the scanner is running in one of your BTs ... it wouldn't if it was completely disabled

[17:08:22]

stuartm:

both BTs in fact

[17:08:30]

skd5aner:

I can double check – I'm actually starting mythtv-setup right now, because I know I have a video source still there I no longer use, so I want to take that out of the equation

[17:08:49]

skd5aner:

I know I've checked in the past, but I could have missed it somehow

[17:11:56]

skd5aner:

stuartm: The EIT scanner options in mythtv-setup general options don't really matter if they're set, correct (they don't need to be "0" or something to disable)

[17:14:35]

stuartm:

under Video Source, 'Perform EIT Scan' should be unchecked if you're not using EIT

[17:15:45]

skd5aner:

under the recorder options, I also have "use DVB card for active EIT scan" unchecked

[17:17:42]

stuartm:

the dvb setting doesn't matter so much as long as the video source setting is unchecked

[17:18:07]

stuartm:

the ones in the 'general' settings again don't matter if it's disabled at the source level

[17:19:02]

skd5aner:

stuartm: just checking everywhere :)

[17:19:11]

stuartm:

stuarta: ^^ we shouldn't be starting EIT scanner threads if EIT scanning is disabled?

[17:20:31]

skd5aner:

stuartm: I have 4 video sources, all of them already had the "Perform EIT Scan" box unchecked...

[17:20:55]

skd5aner:

stuartm: so, if you are seeing EIT scanning active, then I have no idea why as I've already disabled it everywhere I know where to

[17:21:43]

stuartm:

it's not likely to be the cause of your issue, but those threads should never have been started to begin with, they are probably idle

[17:23:22]

skd5aner:

stuartm: ok, well – I'm ok not ruling anything out until it's ruled out :)

[17:23:43]

skd5aner:

stuartm: also, if that thread started and shouldn't have, then perhaps that's a bug?

[17:24:46]

stuartm:

a low severity bug, at worst it ties up threads that my be in limited supply (on some systems)

[17:24:54]

skd5aner:

makes sense

[17:26:23]

skd5aner:

FYI, this server, while not new, is not underpowered... it's a 4 core AMD Athlon II X4 640 Processor

[17:26:33]

skd5aner:

4GB of RAM

[17:26:47]

skd5aner:

just to give you some context

[17:27:54]

skd5aner:

stuartm: I have to run to lunch here in a couple mins, would it be beneficial to open a new ticket specifically for this backtrace to call out the scheduling deadlock? Or attach it to the existing ticket(s)?

[17:28:08]

skd5aner:

I don't want to spam trac if it doesn't make sense to have a different ticket

[17:30:12]

NightMonkey (NightMonkey!~NightrMon@pdpc/supporter/professional/nightmonkey) has quit (Quit: Body blow! Body blow!)

[17:30:40]

skd5aner:

I've got to run, family waiting at the door – I'll check back in later today... thanks again stuartm for all your help!!!!

[17:35:55]

qu (qu!~quassel@c-98-247-207-107.hsd1.wa.comcast.net) has joined #mythtv

[18:25:52]

NightMonkey (NightMonkey!~NightrMon@pdpc/supporter/professional/nightmonkey) has joined #mythtv

[18:42:53]

wylie (wylie!~wylie@ip68-231-120-15.ph.ph.cox.net) has joined #mythtv

[18:50:22]

Steve-Goodey (Steve-Goodey!~steve@host109-158-212-221.range109-158.btcentralplus.com) has quit (Quit: Konversation terminated!)

[18:52:09]

MythBuild:

build #370 of master-win8-msvc-2010–32bit is complete: Failure [4failed Configure and Build] Build details are at http://code.mythtv.org/buildbot/builders/mast . . . t/builds/370 blamelist: Richard Hulme <peper03@mythtv.org >

[18:56:05]

peper03:

Hmm. Why is it still looking in libs for libmythdvdnav? What have I missed?

[19:00:00]

stuartm:

need someone with windows build experience

[19:03:58]

rsiebert (rsiebert!~quassel@g226186162.adsl.alicedsl.de) has quit (Read error: Connection reset by peer)

[19:05:47]

rsiebert (rsiebert!~quassel@f052149068.adsl.alicedsl.de) has joined #mythtv

[19:09:18]

peper03:

Hmm. An entry is missing in external.pro and an explicit cleanup of the VisualStudio project from configure.ps1. I don't see where they're being referenced but hopefully that's it...

[19:12:44]

peper03:

Sorry, I have to leave. I hope that last commit fixes it.

[19:13:20]

peper03:

dblain: If the build fails again, could you let me know what I've done wrong?

[19:39:26]

MythBuild:

build #371 of master-win8-msvc-2010–32bit is complete: Success [3build successful] Build details are at http://code.mythtv.org/buildbot/builders/mast . . . t/builds/371

[19:40:03]

dekarl:

peper03: hmm, splitting whitespace changes and functional changes into separate commits would be nice

[19:40:28]

dekarl:

Oh, looks like you found the same potential solution that I found now ;)

[20:27:30]

doev (doev!~doev@p4FD433AF.dip0.t-ipconnect.de) has quit (Ping timeout: 245 seconds)

[20:28:15]

doev (doev!~doev@p4FD4101E.dip0.t-ipconnect.de) has joined #mythtv

[20:37:25]

caelor:

skd5aner: it's something of a "me too", but your experiences were the same as mine, and I also saw it being ok when I first moved to 0.27, and then got progressively worse

[20:38:25]

caelor:

I "fixed" it by moving my DB to another host (actually running in a Fedora 20 VM, rather than bare metal Ubuntu). I've found the Fedora mysql to be a lot more performant

[20:39:30]

caelor:

it could be a diagnostic step to try. I know Ubuntu has a tendency to have a poor mysql distro config, but there did seem to be something in 0.27's use that progressively reduced its performance further

[20:45:05]

skd5aner:

caelor: thanks – I do take /some/ consolation in that I'm not alone perhaps in this... I've been wondering if I don't have some sort of hardware failure going on or something

[20:45:28]

caelor:

out of curiousity, are you running an ubuntu variant?

[20:45:33]

skd5aner:

although, I've had no indication of such and I did run a bunch of SMART tests last night on my hdds with no obvious errors reported

[20:46:10]

caelor:

another thing that possibly comes to mind is filesystem (type, free space, and possibly fragmentation)

[20:46:19]

skd5aner:

caelor: I am, ubuntu server (not mythbuntu)

[20:46:51]

caelor:

interesting. That's basically what my master backend (and problematic mysql server) is running

[20:47:03]

skd5aner:

well, my OS/DB disk is ext3, all my recording and media disks are xfs

[20:47:37]

caelor:

my DB fs is ext4 (barriers disabled), but likely fragmented as it tends to run on the full side

[20:48:12]

skd5aner:

does barriers impact ext3 too?

[20:48:23]

caelor:

I believe ext3 defaults to disabled barriers

[20:49:03]

caelor:

at the moment, my inclination is to suggest it could be some ubuntu/mysql inefficiency. I'll maybe try and compare my usable fedora mysql config to my ubuntu one

[20:50:17]

skd5aner:

caelor: I do believe that mysql isn't the best performer, but I've tried to follow the most common advice to optimize mysql...

[20:51:13]

caelor:

yes, I had. I was running the config options suggested on the wiki, and had also tried to tune it. With F20, it's just running the wiki tuning suggestions.

[20:51:18]

skd5aner:

but, I'm not a mysql DBA, so it's just what I've stumbled on in the wiki and mailing lists

[20:53:06]

caelor:

yeah. In order to try and isolate root causes, it might be useful to try a clean mysql server, with your existing mythconverg db. It apparently fixed it for me, indicating the fault may lie with mysql/config, rather than mythbackend

[20:54:03]

caelor:

it might also have a chance of improving the usability of your install, until it can be properly resolved

[21:05:40]

skd5aner:

hmmm, I wouldn't be against that idea, that said... I'm not sure I've really tweaked it /that/ much...

[21:05:51]

skd5aner:

but, at this point, I'm open to any suggestion

[21:14:36]

caelor:

well, it worked for me. I didn't think I'd tweaked Ubuntu's mysql much either. It is a concern if we have 2 data points indicating a progressive performance degradation on ubuntu

[21:16:02]

caelor:

it sounded like stuartm's analysis suggested a potential race condition due to slow scheduler (which in my case was likely slow mysql)

[21:28:37]

NightMonkey (NightMonkey!~NightrMon@pdpc/supporter/professional/nightmonkey) has quit (Ping timeout: 248 seconds)

[21:32:40]

dekarl:

hmm, most common advice hopefully is to run this, right? => http://mysqltuner.pl

[21:34:24]

MartinT (MartinT!~smuxi@46-18-104-220.static.vivaciti.org) has joined #mythtv

[21:34:35]

dekarl:

fwiw, I have extensive loggin + database logging + recording to the database disk (only one card with multirec right now, so its not that much) and no real issues... But I did run many rounds of mysqltuner.pl

[21:34:48]

dekarl:

and its the os disk, too

[21:35:09]

dekarl:

and its a plain old spinning rust device, not some funky chip disk

[21:35:18]

caelor:

the changes I had made were the ones recommended in the wiki, followed by mysqltuner recommendations. My DB was on the OS disk, and performance had never been an issue prior to 0.27

[21:36:16]

caelor:

following 0.27, performance seemed better than ever before (for myth), but progressively degraded, until it started causing the issues that skd5aner has also reported. The performance issues affected other databases on mysql too though

[21:36:40]

dekarl:

hmm, I'm on master though

[21:37:01]

caelor:

there were no mysql or other system changes post my 0.27 upgrade (which means I'm on a relatively early 0.27-fixes)

[21:37:08]

dekarl:

how big is your recordedseek and how many recordings do you have?

[21:38:17]

caelor:

moving my DB onto a "clean" mysql running on a Fedora 20 VM (rather than Ubuntu server bare metal) has significantly increased performance (scheduler runs <10s, rather than ~100–200s)

[21:39:44]

dekarl:

I have 16,204,881 entries in recorded seek for 1001 recordings

[21:40:32]

dekarl:

I vaguely remember a bug where we would write 2 entries per frame instead of 2 per keyframe, need to look that up to see if I remeber right...

[21:40:34]

caelor:

recordedseek was 1305936 rows, 74.2Mb. 260 recordings.

[21:40:45]

dekarl:

hm, sounds about right

[21:41:09]

caelor:

Given the problem went away with moving to a F20 Mysql server, I was suspecting something Ubuntu.

[21:42:48]

caelor:

the db on the F20 machine was copied using the db backup scripts, chaning DB config to point to the new server, and then restoring the backup using the scripts, so should have been like for like

[21:43:59]

dekarl:

so you switched from mysql to mariadb?

[21:44:08]

caelor:

arguably, yes.

[21:44:20]

NightMonkey (NightMonkey!~NightrMon@pdpc/supporter/professional/nightmonkey) has joined #mythtv

[21:44:22]

caelor:

:) I forgot about that

[21:45:21]

caelor:

if there was truly an issue with Ubuntu config mysql, I would have expected more widespread reports, so there must be some other common element between skd5aner and myself

[21:46:48]

stuartm:

fwiw, we shouldn't end up deadlocking even if the scheduler takes hours, so there is a bug to be fixed there somewhere, but we're still going to have problems elsewhere if the scheduler does take a while to run and so we need to look at any performance issues, and also reduce the number of places which have to wait for the scheduler to complete

[21:46:54]

dekarl:

which version of ubuntu was that? 12.04?

[21:47:58]

caelor:

12.04, installed from Ubuntu Server. 32bit install. mysql Ver 14.14 Distrib 5.5.35

[21:51:24]

dekarl:

caelor: I'm out of idea...

[21:52:52]

caelor:

yes. For me the problem went away with using the F20 server (and I'm planning to move the master backend onto Fedora in the upcoming weeks anyway), but I think skd5aner is still seeing the issue.

[21:53:36]

caelor:

I'll keep the Ubuntu box around for a while, so should be able to check for "suspected bad" config options, if there's more investigation

[21:54:18]

dekarl:

stuartm: you don't have a Language in your settings table for the host where you ran sendProfile.py?

[21:54:56]

stuartm:

dekarl: I do

[21:55:05]

stuartm:

en_GB

[21:55:16]

stuartm:

it's global

[21:55:24]

stuartm:

so host is NULL

[21:55:39]

B34N (B34N!~mark@pool-72-82-168-153.cmdnnj.east.verizon.net) has joined #mythtv

[21:55:54]

stuartm:

hmm, wonder if the python bindings are handling global/local settings correctly

[21:56:07]

B34N (B34N!~mark@pool-72-82-168-153.cmdnnj.east.verizon.net) has left #mythtv ()

[21:56:20]

dekarl:

strange, "self._data.language Â Â Â = _SETTINGS.Language.lower()" giving "AttributeError: 'NoneType' object has no attribute 'lower'" hints that "_SETTINGS = _DB.settings[_DB.gethostname()]" did not fallback to the global setting

[21:56:40]

stuartm:

should always check local then global and only return nothing if neither exist

[21:57:26]

stuartm:

dekarl: it should be global for every new install now, I made that change a few weeks back

[22:08:02]

dekarl:

stuartm: I changed my Language setting to be global, too. No I get the same backtrace

[22:08:08]

dekarl:

s/No /Now /

[22:16:45]

dekarl:

This hack worksforme in all three cases, http://pastebin.com/U7z07NMd but I have never done Python, so I'd appreciate a peer review before I commit that

[22:21:22]

stuartm:

looks reasonable, but I've only dabbled in python, wagnerrp should probably be the one to take a look

[22:22:23]

stuartm:

TBH, the python bindings should probably have a GetSetting() that does the right thing

[22:34:39]

dekarl:

I'm not finding such a function... IT should be good enough to move on to the next thing that keeps Smolt from returning Mageia

[22:39:43]

MartinT:

can anyone tell what smolt and mageia are, and what they do in this context... it sounds like an interesting conversation, but no idea what it's about...

[22:40:30]

dekarl:

Mageia being a Linux distribution and Smolt being a tool to get a rough idea of our installed base http://smolt.mythtv.org/

[22:41:00]

dekarl:

notice that the OS overview on our Smolt does not show Mageia, I'm wondering why that is so

[22:43:13]

skd5aner:

dekarl: I've ran mysqltuner.pl – nothing stood out, everything was within the norms

[22:45:10]

dekarl:

skd5aner: which suggests that you found all needed configuration parameters already. When I ran it on plain 12.04LTS I had to do like 5–10 rounds to get sensible numbers

[22:48:11]

MartinT:

2 installs with 16 cpus... jeez...

[22:49:44]

dekarl:

I bet its "only" 2x quad core with Hype Threading.

[22:49:55]

MartinT:

still...

[22:50:00]

doev (doev!~doev@p4FD4101E.dip0.t-ipconnect.de) has quit (Read error: Operation timed out)

[22:51:20]

peper03:

dekarl: Sorry, whitespace changes were unintentional. Seems the editor stripped trailing whitespace automatically.

[22:51:49]

dekarl:

no worries, just made me wonder what you changed in that lines ;)

[22:56:31]

Steve-Goodey (Steve-Goodey!~steve@host109-158-212-221.range109-158.btcentralplus.com) has joined #mythtv

[22:58:22]

dekarl:

stuartm: can you try again and pastebin the result? I get country "none" though after changing that to a global setting, but at least its not dieing.

[23:06:35]

stuartm:

note that Smolt is an opt-in feature, so at best it gives us a guide

[23:06:48]

stuartm:

dekarl: sure, hang on a mo

[23:07:15]

skd5aner:

dekarl: myrecordseek table has 10,522,394 rows and recorded table has 2091 rows

[23:07:23]

stichnot (stichnot!~stichnot@mythtv/developer/stichnot) has joined #mythtv

[23:10:03]

skd5aner:

I can tell you the absolute WORSE behavior is when mythfilldatabase is running...

[23:10:04]

dekarl:

stuartm: I know its opt-in, and it depends on Python bindings (hi MacOSX and Windows), so has to be taken with a grain of salt. But I'd prefer it actually worked if someone was so nice to opt in :D

[23:10:10]

skd5aner:

when it's running myth doesn't want to do anything

[23:10:41]

skd5aner:

I'm pulling a backtrace now from mythbackend when mythfilldatabase is running. Maybe I should get one from mythfilldatabase too

[23:11:35]

dekarl:

skd5aner: looking into iotop might give more insight... possibly just mysql trashing due to temporary (sorting/joining) tables hitting the disc...

[23:12:52]

skd5aner:

dekarl: ok, not really familiar with iotop... I'll give it a go

[23:13:38]

stuartm:

OS: Mageia 3 thornicroft

[23:14:06]

skd5aner:

stuartm: do you think gdb output of mbe and mfdb would be worth looking at?

[23:15:28]

stuartm:

dekarl: oh I agree, just clarifying for others present that when you say "installed base" you don't mean "number of MythTV users"

[23:16:41]

dekarl:

ohh, I see. Yes, its just a random sample of installations that have opted in and don't run into issues trying to run Smolt :/

[23:18:06]

skd5aner:

does mythbuntu automatically opt-in for the users, or make it very easy for them to do? I ask because the smolt data has always seemed highly propotiante to mythbuntu users, and I didn't know if that's because it's literally that high of a ratio, or if they make it easy (or always opt-in) for their users to submit a smolt profile

[23:19:20]

skd5aner:

s/proprtionate/disproportionate

[23:23:23]

skd5aner:

I have to admit, I've considered buying an SSD for the OS drive – as much of a bandaid as it might be...

[23:25:07]

dekarl:

http://code.mythtv.org/cgit/mythtv/tree/mytht . . . file.cpp#n26 looking for that setting and the earlier variant does not show a default around the deb packaging

[23:29:09]

Steve-Goodey:

skd5aner: I've just done a clean mythbuntu install. It's in the frontend, setup, setup wizard. It offers to submit, view delete your hardware profile.

[23:29:10]

skd5aner:

I'm (embarrassing) running my OS/DB on an older 200GB PATA drive... (yes, you heard that right)

[23:30:06]

skd5aner:

that said, it's ran like a tank for 8+ years and doesn't show any current signs in SMART that it has any issues

[23:30:58]

skd5aner:

Steve-Goodey: actually, I did the same thing about a week ago as I built a new frontend, and for the first time I actually used mythbuntu rather than building myself...

[23:31:37]

skd5aner:

Steve-Goodey: and yes, I saw it the same way, via the setup wizard – but, I always wondered if they had enabled it automatically or people who didn't go through the wizard

[23:31:50]

skd5aner:

s/or/for...

[23:31:57]

skd5aner:

laptop keys are getting flaky :/

[23:32:05]

Steve-Goodey:

First time I've used an SSD so far so good.

[23:32:44]

skd5aner:

yea, I put a 60GB in my new frontend box... it was overkill for sure, but it's for a frontend in a bedroom, so I really wanted it for noise reduction more than anything

[23:33:57]

stichnot (stichnot!~stichnot@mythtv/developer/stichnot) has quit (Ping timeout: 248 seconds)

[23:34:40]

MartinT:

skd5aner: my slave be is using a 160GB IDE...

[23:35:22]

skd5aner:

technically, my slave/fe is using an IDE drive too, although it's a laptop drive and it doesn't store anything on it other than the OS and binaries... no media, no recordings, no DB

[23:35:24]

MartinT:

master is in a VM, but that's only running off a 160GB RAID 1 (or 0 whichever is mirror)

[23:35:31]

skd5aner:

1 is mirror

[23:35:33]

skd5aner:

0 is stripe

[23:36:08]

MartinT:

always get the 2 mixed up

[23:36:55]

skd5aner:

I have to deal with RAID levels on a daily basis, so yea... although sometimes I don't have to deal with the more exotic flavors and have to look them up on occasion

[23:37:33]

MartinT:

master be is a quad core host (ESXi) with 3 vms... the slave is a 2.8GHz pentium... running for about 3 years... no signs of failure...

[23:37:54]

MartinT:

when it goes, I've got 2.8GHz Meridian box waiting to take over...

[23:38:13]

MartinT:

then a 2.4GHz Dell if I get desperate

[23:38:31]

MartinT:

but anyway... all of them using Mythbuntu...

[23:39:01]

MartinT:

although the setup process could be made a massive amount more slick... it does work well

[23:39:09]

Steve-Goodey (Steve-Goodey!~steve@host109-158-212-221.range109-158.btcentralplus.com) has quit (Quit: Konversation terminated!)

[23:39:23]

MartinT:

mythbuntu just makes it that little bit easier...

[23:39:36]

MartinT:

anyway, off to bed...

[23:39:48]

MartinT (MartinT!~smuxi@46-18-104-220.static.vivaciti.org) has quit (Read error: Connection reset by peer)

[23:40:46]

skd5aner:

I have to admit – I was pretty impressed with the amount of stuff I didn't have to do to get mythbuntu working that I would have had to do manually if I would have just deployed vanilla ubuntu server

[23:43:05]

skd5aner:

if I don't have to touch lirc, I consider that a success

[23:46:58]

Tobbe5178 (Tobbe5178!~asdf@h104n2-sv-a13.ias.bredband.telia.com) has quit (Read error: Connection reset by peer)

[23:48:45]

dekarl:

wagnerrp: If you get a change to look at the two small Smolt fixes I just commited that would be appreciated. Looks like new installs of master/0.27 could not send the profile since late August. (old installs worked)

[23:48:55]

skd5aner:

dekarl: ran iotop, here was the output -http://pastebin.com/TC0pnnTU

[23:49:48]

stuartm:

it's a great shame that the native input work was ultimately hamstrung by X11 :(

[23:50:37]

stuartm:

if it hadn't been, then no-one would need lirc, remotes would just work

[23:51:20]

dekarl:

oh, that would be nice... mythtv being able to get the "Volume Up/Down" keys natively

[23:52:04]

stuartm:

just need for X11 to be put to death

[23:52:05]

dekarl:

skd5aner: that is iotop without mythfilldatabase, right? looks like a quiet time

[23:52:26]

dekarl:

did anybody try mythfrontend on qt on directfb?

[23:52:44]

skd5aner:

dekarl: mythfilldatabase is still running

[23:52:55]

skd5aner:

actually, JUST quit

[23:53:34]

skd5aner:

ran from 16:05 to 18:52

[23:54:26]

stuartm:

dekarl: not sure if that would work, the problem is the keypress handling of the X11 protocol, they allow an 8 bit int, which only covers the keys on most keyboards and a few media keys (but not all of them)

[23:55:11]

stuartm:

don't know how qt handles input when using directfb

[23:55:31]

stuartm:

skd5aner: 3 hours?!

[23:55:50]

skd5aner:

yup

[23:55:57]

stuartm:

4 hours!

[23:56:07]

skd5aner:

nearly, yup

[23:56:24]

stuartm:

that's insane – this is SD?

[23:56:36]

skd5aner:

yes, 3 sources

[23:56:42]

skd5aner:

--dd-grab-all

[23:57:05]

dekarl:

ohh, I was wondering what could be wrong with 4 minutes...

[23:57:51]

stuartm:

no, three hours, I was right the first time (it's late, I'm tired)

[23:57:59]

skd5aner:

stuartm: http://pastebin.com/p3agkfN5 – not verbose, but you can see the long waits

[23:58:07]

dekarl:

stuartm it seems QT uses the input device abstraction from directfb (which was written for STB, so it might actually "just work")

[23:58:09]

skd5aner:

yea, sorry... 3 hours :)

[23:58:57]

dekarl:

what does mysqltuner.pl say now?

[23:58:58]

stuartm:

skd5aner: there's something really wrong with your database

[23:59:16]

stuartm:

2014-01–25 16:49:02.400722 I Clearing from 2014-01–23T19:00:00 to 2014-02–09T18:59:59 (localtime)

[23:59:20]

stuartm:

2014-01–25 17:52:34.229347 I Data for source cleared.

[23:59:39]

skd5aner:

dekarl: when mfdb isn't running?

[23:59:48]

stuartm:

an hour to DELETE FROM the program table?

MythLogBot@irc.freenode.net :: #mythtv

Current users (84):