IRC Logs for #circuits Wednesday, 2015-08-12

prologicRomster, hi :)00:25
prologicI investigated this last night00:25
prologicI did not find any anomaly in the event flows00:25
prologicso I'm not sure wtf's going on there00:25
kdbUptime: 0+10:21:36 (CPU: 118.24s 0.32%)00:25
kdbUptime: 0+10:21:42 (CPU: 118.27s 0.32%)00:26
prologicreally weird00:26
*** kdb has quit IRC00:28
*** kdb has joined #circuits00:28
*** kdb has quit IRC00:28
*** kdb has joined #circuits00:28
prologickdb, uptime00:29
kdbUptime: 0+0:0:49 (CPU: 0.48s 0.97%)00:29
kdbUptime: 0+0:0:58 (CPU: 0.52s 0.88%)00:29
kdbUptime: 0+0:1:22 (CPU: 0.60s 0.73%)00:30
kdbUptime: 0+0:1:23 (CPU: 0.61s 0.73%)00:30
kdbUptime: 0+0:1:25 (CPU: 0.62s 0.73%)00:30
prologic[00:34:29] <prologic> @uptime00:30
prologic[00:34:29] <prologic> kdb, uptime00:30
prologicbit of a strange behavior there00:30
prologicseems that broadcast rightly translates @<command> into <bot>, <command00:30
prologicbut the logging plugin picks this up :)00:30
prologickdb, uptime00:30
prologicok so see00:31
prologicthat time it didn't respond00:31
prologicbut that line got logged00:31
prologic[00:35:18] <prologic> kdb, uptime00:31
prologic[00:35:27] <prologic> ok so see00:31
prologic[00:35:29] <prologic> that time it didn't respond00:31
prologic[00:35:33] <prologic> but that line got logged00:31
prologicthe bot is definitely getting the requests00:31
prologicand it's definitely trying to response00:32
prologicI can see it in the event logs00:32
prologicthis is gonna bug me till I figure this out00:32
*** kdb has quit IRC00:37
*** kdb has joined #circuits00:38
*** kdb has quit IRC00:38
*** kdb has joined #circuits00:38
*** kdb has quit IRC00:52
*** kdb has joined #circuits00:54
*** kdb has quit IRC00:54
*** kdb has joined #circuits00:54
prologicRomster, ping?00:55
prologiccould you try to reproduce this weird issue with the kdb that's here now?00:55
prologicI'm running it non-Docker on my desktop instead00:55
prologicto see00:55
prologicI reproduced it just then00:55
Romsterwas at work02:47
Romsteroff again03:11
prologic<_read[bot] (<socket._socketobject object at 0x7f4becabaf30> )>03:37
prologic<read[bot] (':prologic!~prologic@unaffiliated/prologic PRIVMSG kdb :uptime\r\n' )>03:37
prologic<line[bot] (':prologic!~prologic@unaffiliated/prologic PRIVMSG kdb :uptime' )>03:37
prologic<privmsg[bot] ((u'prologic', u'~prologic', u'unaffiliated/prologic'), u'kdb', u'uptime' )>03:37
prologic<uptime[commands] ((u'prologic', u'~prologic', u'unaffiliated/prologic'), u'prologic', '' )>03:37
prologic<uptime_done[commands] ('Uptime: 0+2:40:9 (CPU: 29.86s 0.31%)' )>03:37
prologic<request[bot] ("PRIVMSG prologic :Uptime: 0+2:40:9 (CPU: 29.86s 0.31%)" )>03:37
prologic<write[bot] ('PRIVMSG prologic :Uptime: 0+2:40:9 (CPU: 29.86s 0.31%)\r\n' )>03:37
prologic<_write[bot] (<socket._socketobject object at 0x7f4becabaf30> )>03:37
prologicanyone got any thoughts on this?03:37
prologicas to why the response would never come back?03:38
prologicfrom the above event flows03:38
prologicthe response is being written back to the socket just fine03:38
prologicand appears no different to any time it *does* work03:38
*** kdb has quit IRC06:48
*** Osso has joined #circuits07:49
*** ninkotech__ has quit IRC08:39
prologichey Osso08:43
prologicwondering what your thoughts on on this:
prologicreally weird behavior we've been seeing08:51
*** kdb has joined #circuits08:55
*** kdb has quit IRC08:55
*** kdb has joined #circuits08:55
*** Coldblackice has quit IRC08:55
*** Coldblackice has joined #circuits09:00
*** kdb has quit IRC09:00
*** kdb has joined #circuits09:02
*** kdb has quit IRC09:02
*** kdb has joined #circuits09:02
spaceoneback since yesterday10:11
spaceonei just opened
spaceonecan you help me with this?10:11
prologicYeah I sure can10:12
prologiccan you help with this memory leak issue? :)10:13
spaceonewell, lets talk about the cause first10:13
spaceonejah, but i dont fully understand the memory leak10:13
spaceonethe socket error doesn't happen when i connect to gooogle.com10:13
spaceoneprobably because the OS CA's know google10:14
spaceonecan i fix it by specifying a certfile?10:14
prologicjust ran your sample10:14
prologicand get an endless _read events10:14
prologicyeah hold on a sec10:15
prologicisn't this issue related to10:15
prologiclemme find it10:15
prologicI think fixing #114 will fix #11510:15
spaceonei saw this, too10:15
prologicthis is all about ssl10:15
prologicwe basically don't do this properly10:15
prologicat all10:15
prologicwe have to confirm the ssl handshake10:15
prologicbefore firing the connected event10:16
prologicotherwise the socket just isn't valid nor ready10:16
spaceoneprologic: but there is not a 'connected' event fired at all yet!10:18
spaceone(the exception causes that none is fired)10:18
spaceonebut the socket is added to the poller10:18
spaceoneand then self._sock gets overwritten10:19
spaceoneso _close() will discard the wrong socket?10:19
spaceonesee the last comment10:23
spaceoneso, now i don't understand the SSLError: [SSL: UNKNOWN_PROTOCOL] unknown protocol (_ssl.c:581)10:25
spaceoneand if i specify a certfile i get SSLError: [SSL] PEM lib (_ssl.c:2525)10:26
prologici think more investigation is needed10:28
prologicI did find some code where proper async handshanke is done10:28
prologicI was going to see if I could borrow that10:28
prologicbut I have to go now; so we'll continue this tmorrow?10:28
*** ninkotech_ has joined #circuits13:27
*** Coldblackice has quit IRC16:16
*** Coldblackice has joined #circuits21:53
prologicspaceone: Let's develop a 3.2 milestone23:41
prologicand work out what things we have left that we *should* do23:41
prologicI want to also incorporate
prologicas part of packaging/distribution23:42
prologicso we don't have to manually manage the version anymore23:42
prologicI renamed "next" to "3.2"23:46
prologicthat should do23:46
prologicbut we should review what's there and possibly move some things out to a 3.2.1 or 3.3 or so23:46
prologicI'd like us to get 3.2 out the door soon23:46
prologicAlso I think we need to revisit namespaces again23:47
prologicand work out whether it's worth doing23:47
prologicSee my comments on the issue re Flask's decision not to adopt namespaces23:47
prologicI'll also look at fixing the SSL Handshake sync issue tonight23:47
prologicOsso, Did you get time yet to look at this memory leak(s)?23:47

Generated by 2.11.0 by Marius Gedminas - find it at!