| 00:34.41 | *** join/#asterisk-dev serafie (~erin@user-24-214-173-250.knology.net) |
| 02:19.20 | *** join/#asterisk-dev serafie (~erin@user-24-214-173-250.knology.net) |
| 02:43.02 | *** join/#asterisk-dev pabelanger (~pabelange@asterisk/contributor-and-bug-marshal/pabelanger) |
| 02:43.02 | *** mode/#asterisk-dev [+o pabelanger] by ChanServ |
| 05:29.34 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.126.21) |
| 06:46.32 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.126.236) |
| 06:52.04 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.126.236) |
| 07:16.41 | *** join/#asterisk-dev L0c0 (~nde@host-212-68-194-46.brutele.be) |
| 07:44.55 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.127.66) |
| 07:49.03 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.127.66) |
| 08:10.39 | *** join/#asterisk-dev GameGamer43 (uid5533@gateway/web/irccloud.com/x-pxywtayexddicpuf) |
| 08:13.31 | *** join/#asterisk-dev tonyclewis (uid6025@gateway/web/irccloud.com/x-tnetocevdapqvbqm) |
| 09:11.06 | *** join/#asterisk-dev vlad_starkov (~vlad_star@195.68.167.197) |
| 09:57.01 | *** join/#asterisk-dev hehol (~hehol@2001:1438:1009:200:b1b3:2b0a:9ba9:f6f1) |
| 10:08.24 | *** join/#asterisk-dev elguero (~miguel323@2001:470:1f06:12c4::2) |
| 11:30.51 | *** join/#asterisk-dev oej (~oej@2001:df8:1:64:58f4:a3e5:8345:f8ca) |
| 11:30.52 | *** mode/#asterisk-dev [+o oej] by ChanServ |
| 11:44.35 | *** join/#asterisk-dev acidfu (~nib@unaffiliated/acidmen) |
| 11:52.22 | *** join/#asterisk-dev w9sh (~sph@50-79-224-193-static.hfc.comcastbusiness.net) |
| 12:08.43 | *** join/#asterisk-dev pabelanger (~pabelange@asterisk/contributor-and-bug-marshal/pabelanger) |
| 12:08.44 | *** mode/#asterisk-dev [+o pabelanger] by ChanServ |
| 12:15.31 | *** join/#asterisk-dev vlad_starkov (~vlad_star@195.68.167.197) |
| 12:39.05 | *** join/#asterisk-dev pabelanger (~pabelange@asterisk/contributor-and-bug-marshal/pabelanger) |
| 12:39.05 | *** mode/#asterisk-dev [+o pabelanger] by ChanServ |
| 13:11.55 | *** join/#asterisk-dev putnopvut (~putnopvut@asterisk/master-of-queues/mmichelson) |
| 13:11.55 | *** mode/#asterisk-dev [+o putnopvut] by ChanServ |
| 13:35.34 | *** join/#asterisk-dev vlad_starkov (~vlad_star@195.68.167.197) |
| 13:43.43 | *** join/#asterisk-dev mjordan (~mjordan@nat/digium/x-bjbrjdxprqdvclzh) |
| 13:43.43 | *** mode/#asterisk-dev [+o mjordan] by ChanServ |
| 13:51.26 | *** join/#asterisk-dev serafie (~erin@nat/digium/x-gqmwqlwmthzswypn) |
| 13:51.48 | putnopvut | Welp, color me baffled. Trying to merge in the removal of ast_bridged_channel, and all of a sudden chan_misdn doesn't compile due to an implicit declaration of function "ast_pickup_call". I noticed that pickup.h wasn't included, so I now #include "asterisk/pickup.h" in chan_misdn.c. Same compile error. make clean and make distclean aren't helping |
| 13:52.15 | file | I believe mjordan merged in changes re: that a short period of time ago |
| 13:52.55 | putnopvut | I think I see the problem |
| 13:53.03 | putnopvut | pickup.h and features.h use the same #ifndef name |
| 13:53.15 | putnopvut | file: had done an svn up just before. |
| 13:53.42 | file | nods |
| 13:54.54 | putnopvut | Let's see if that worked. |
| 13:55.13 | putnopvut | Success! |
| 13:55.21 | file | gold star! |
| 14:02.34 | mjordan | blech |
| 14:02.38 | mjordan | sorry everyone. |
| 14:03.00 | mjordan | I think I might just leave the land o renaming and move on |
| 14:03.13 | file | to greener pastures? |
| 14:03.16 | mjordan | I hope |
| 14:05.05 | file | I mayyyy create a branch for experimental SIP stuff in the future... for kicks |
| 14:05.58 | file | talks to self |
| 14:08.33 | *** join/#asterisk-dev tzafrir (~tzafrir@bzq-218-28-58.cablep.bezeqint.net) |
| 14:08.33 | *** mode/#asterisk-dev [+o tzafrir] by ChanServ |
| 14:10.28 | *** join/#asterisk-dev kharwell (~kharwell@nat/digium/x-mykgywmfcjogrilk) |
| 14:21.28 | mjordan | damn you snmp! |
| 14:24.26 | *** join/#asterisk-dev pabelanger (~pabelange@asterisk/contributor-and-bug-marshal/pabelanger) |
| 14:24.26 | *** mode/#asterisk-dev [+o pabelanger] by ChanServ |
| 14:28.48 | *** join/#asterisk-dev italorossi (~italoross@187.60.66.11) |
| 14:45.59 | leifmadsen | mjordan: o.O |
| 14:47.03 | mjordan | leifmadsen: snmp is currently breaking the build. |
| 14:47.12 | leifmadsen | sounds right :) |
| 14:47.12 | mjordan | snmp, not the code change :-P |
| 14:51.30 | *** join/#asterisk-dev pabelanger (~pabelange@asterisk/contributor-and-bug-marshal/pabelanger) |
| 14:51.30 | *** mode/#asterisk-dev [+o pabelanger] by ChanServ |
| 14:58.50 | *** part/#asterisk-dev leedm777 (~leedm777@nat/digium/x-odgoteiyrrpcojqe) |
| 15:00.22 | wdoekes | mjordan: some of r374177 from 1.8 didn't go into 10 in r374178, but went into 11 in r374196, like the change to res_musiconhold.c. (yes.. I know 10 is EOL, but this was october last year) |
| 15:01.52 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.125.212) |
| 15:13.56 | WIMPy | putnopvut: I tried your suggestion and commented out ast_channel_publish_varset(). That doesn't seem to have a significant impact. |
| 15:17.55 | putnopvut | very interesting |
| 15:34.41 | *** join/#asterisk-dev vlad_starkov (~vlad_star@93.191.18.82) |
| 15:57.17 | WIMPy | I tried a little more often. There is a slight improvement, but it's nowhere near the change between versions. |
| 15:58.49 | WIMPy | Any other suggentions what I could try, anyone? |
| 15:59.17 | *** join/#asterisk-dev leedm777 (~leedm777@nat/digium/x-odgoteiyrrpcojqe) |
| 16:02.36 | *** join/#asterisk-dev anonymouz666 (~anonymouz@189-25-41-60.user.veloxzone.com.br) |
| 16:08.54 | mjordan | wdoekes: do you think it's worth putting a patch into 10 for that at this point? |
| 16:53.28 | *** join/#asterisk-dev vlad_sta_ (~vlad_star@109.188.125.212) |
| 17:04.28 | coreyf1513 | WIMPy: have you tried using a profiler? |
| 17:04.57 | WIMPy | Nope. Haven;t played with these things at all so far. |
| 17:11.44 | coreyf1513 | WIMPy: 'valgrind --tool=callgrind asterisk' works on asterisk. I could not get gprof to work, and never tried oprofiled. |
| 17:11.48 | *** join/#asterisk-dev [TK]D-Fender (~chatzilla@64.235.216.2) |
| 17:36.50 | WIMPy | valgrind segfaults :-( |
| 17:38.38 | coreyf1513 | valgrind has effects on timing, the only thing I've seen that cause is deadlocks in a couple weird places.. did you get a core dump? |
| 17:39.21 | WIMPy | I try to re-build without optimizations. |
| 17:39.53 | WIMPy | That was better. |
| 17:41.43 | WIMPy | Erm, I guess I need to start asterisk with -c. |
| 17:43.02 | WIMPy | Wow. That is slooooow. |
| 17:45.12 | coreyf1513 | yeah it might be good to disable module autoload and load only whats needed.. |
| 17:45.33 | WIMPy | I see that. |
| 17:45.52 | coreyf1513 | also because the callgrind will produce results for the duration of the process, including startup/shutdown |
| 17:46.38 | WIMPy | Yes, it's running for two minutes now and still loading. |
| 17:47.55 | WIMPy | Err, actually the timestamps the *CLI outpus don't seem to be in realtime any more. Interesting. |
| 17:48.03 | coreyf1513 | yeah |
| 17:48.39 | coreyf1513 | valgrind seems to reduce the amount of thread switching.. I know the logger thread falls behind |
| 18:02.35 | *** join/#asterisk-dev pabelanger (~pabelange@asterisk/contributor-and-bug-marshal/pabelanger) |
| 18:02.35 | *** mode/#asterisk-dev [+o pabelanger] by ChanServ |
| 18:08.38 | WIMPy | Interesting. |
| 18:10.00 | WIMPy | Although the test takes forever with valgrind, I do not end up with a stuck channel. |
| 18:10.09 | WIMPy | Well, not on the first try... |
| 18:11.46 | coreyf1513 | probably due to the reduced thread switching. I don't know what thread would decide when a channel thread is frozen, but I'm guessing that it's running less frequently when you use valgrind |
| 18:14.16 | WIMPy | Ok, what do I make of that callgrind.out? Is there somethign to parse that? |
| 18:14.32 | opticron | file, looking at the leedm777's comments on https://reviewboard.asterisk.org/r/2704/, I know that sorcery doesn't have an equivalent for the config framework's aco_set_defaults(), but I'm trying to detect the lack of a config section, instantiate the appropriate object with defaults, and apply it |
| 18:14.36 | coreyf1513 | callgrind_annotate |
| 18:14.37 | opticron | I'm failing miserably at the moment |
| 18:15.15 | opticron | I'm currently attempting this with ast_sorcery_alloc and ast_sorcery_create |
| 18:15.32 | file | _create creates the object in a wizard |
| 18:16.16 | WIMPy | coreyf1513: Do I get that right, that it only lists the number of calls? No timing, right? |
| 18:16.21 | file | ast_sorcery_alloc automatically applies defaults, so you'd have to store the returned object somewhere yourself |
| 18:17.22 | coreyf1513 | WIMPy: not really, it's the number of polling cycles that asterisk was in each procedure |
| 18:18.15 | coreyf1513 | WIMPy: so 1,000,000 calls to a slow procedure will give a higher number than 1,000,000 calls to a fast procedure |
| 18:18.45 | WIMPy | Doesn't really tell me much :-( I added it to http://wimpy.yeti.dk/pastebin |
| 18:19.36 | WIMPy | .. in the hope that someone can make more sense of it. |
| 18:23.05 | coreyf1513 | you would need separate runs - one with SIP, one with IAX. or SIP on 11 and SIP on trunk. that way the results can be compared |
| 18:23.38 | WIMPy | Ok. |
| 18:24.17 | WIMPy | I used LCR as it doesn't do anything else, like SIP and IAX. |
| 18:34.49 | mjordan | it is spending more time than I would expect resolving XPath arguments. |
| 18:34.50 | WIMPy | Even with heavily increased timeout, I'm unable to get an iax call through. |
| 18:35.42 | mjordan | WIMPy: looks like you're running with DEBUG_THREADS enabled too |
| 18:36.07 | WIMPy | All debuggin on. |
| 18:36.18 | coreyf1513 | I was just going to say that.. turning off all debug options will give a cleaner profile |
| 18:36.25 | mjordan | DEBUG_THREADS is a KILLER. |
| 18:36.26 | file | that will *kill* your performance |
| 18:36.45 | mjordan | and may actually make trunk run worse than previous versions, given how things have moved around in ye ol' concurrency model |
| 18:36.51 | mjordan | DEBUG_THREADS basically single threads all of Asterisk |
| 18:37.09 | WIMPy | file: It didn't do so until two weeks or so ago |
| 18:37.14 | mjordan | so all of the work that was done to put things into thread pools and push reporting behavior off of the pbx threads is actually now worse than it used to |
| 18:38.38 | WIMPy | Ok, so it's the debugging that has become a killer? |
| 18:38.55 | coreyf1513 | debugging has always been a performance killer |
| 18:39.14 | file | the degree to which it kills has changed due to a fundamental shift in how things are done |
| 18:39.17 | mjordan | DEBUG_THREADS always has been a killer, and what it does to your system may actually make it impact performance more in later versions of Asterisk. That trend has continued on into trunk |
| 18:40.10 | WIMPy | Ok. |
| 18:40.19 | WIMPy | So anyone interested in sip results anyway? |
| 18:43.43 | mjordan | sure, I'm looking at the pb |
| 18:44.00 | WIMPy | Ok, will do some then. |
| 18:45.19 | coreyf1513 | the new stuff on pb references the same file 'callgrind.out.21821' ? also the results don't seem to add up |
| 18:46.02 | WIMPy | Wait. |
| 18:47.41 | WIMPy | http://wimpy.yeti.dk/callgrind.lcr.inclusive=no http://wimpy.yeti.dk/callgrind.lcr.inclusive=yes |
| 18:47.46 | WIMPy | http://wimpy.yeti.dk/callgrind.sip.inclusive=no http://wimpy.yeti.dk/callgrind.sip.inclusive=yes |
| 18:53.26 | *** join/#asterisk-dev eZz (~ez@46.164.133.77) |
| 18:53.58 | WIMPy | scratches is head |
| 18:54.19 | WIMPy | Ok, so DEBUG_THREADS is indeed the killer. |
| 18:54.50 | WIMPy | But the only thing I did was to disable that and recompile. No I can't ser verbose down in the remote console any more. |
| 18:54.58 | WIMPy | WTF??? |
| 18:55.12 | WIMPy | *Now |
| 18:57.24 | WIMPy | Dang. I remember I once had the issue that I couldn't set the remote console vverbose level higher than what was set in asterisk.conf. |
| 18:57.36 | WIMPy | Now I can't set it lower. |
| 18:57.41 | WIMPy | But I can set it higher. |
| 18:59.54 | coreyf1513 | WIMPy: are you sure the SIP test actually ran? I don't see any hits for pbx_exec, pbx_builtin_setvar, __ast_pbx_run. they all appear in the LCR test output |
| 19:00.23 | WIMPy | yes, I did see the loop executing in the *CLI. |
| 19:01.21 | WIMPy | Well, We have a saying here that fits the situation perfectely. |
| 19:02.03 | WIMPy | It works because of the similarity of the words to measure and rubbish. |
| 19:02.12 | WIMPy | "Wer misst misst mist". |
| 19:02.27 | WIMPy | Who measures measures rubbish. |
| 19:02.39 | WIMPy | True story. |
| 19:16.12 | *** join/#asterisk-dev help (andrew@wallace.mixdown.ca) |
| 19:20.23 | *** join/#asterisk-dev tzafrir_laptop (~tzafrir@bzq-218-28-58.cablep.bezeqint.net) |
| 19:20.23 | *** mode/#asterisk-dev [+o tzafrir_laptop] by ChanServ |
| 19:22.37 | WIMPy | Ok, so the debugging was also a good explanation for different timing using different channeltypes, I thought. But even without I see a difference. |
| 19:40.43 | *** join/#asterisk-dev wdoekes (~walter@wjd.osso.nl) |
| 19:57.54 | *** join/#asterisk-dev w9sh (~sph@50-79-224-193-static.hfc.comcastbusiness.net) |
| 20:00.08 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.124.254) |
| 20:21.37 | coreyf1513 | WIMPy: what is the difference between channel types with debugging off? is there still a big difference between versions? |
| 20:22.39 | WIMPy | I have only tested the current version. But sip is obviousely slower. |
| 20:22.51 | WIMPy | (again) |
| 20:24.03 | WIMPy | Unfortunately the impact ot DEBUG_THREADS on real calls was quite little compared to the test case. |
| 20:27.52 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.124.27) |
| 20:28.54 | *** join/#asterisk-dev vlad_starkov (~vlad_star@109.188.124.27) |
| 21:05.03 | leifmadsen | https://github.com/mrts/git-svn-bridge |
| 21:05.04 | leifmadsen | that. |
| 21:17.10 | *** join/#asterisk-dev tonyclewis (uid6025@gateway/web/irccloud.com/x-nefworoqpwenfjwy) |
| 21:50.39 | *** part/#asterisk-dev mjordan (~mjordan@nat/digium/x-bjbrjdxprqdvclzh) |
| 22:12.24 | *** part/#asterisk-dev kharwell (~kharwell@nat/digium/x-mykgywmfcjogrilk) |