| Summary: | ASTERISK-01345: LibPri dies after about 90k calls | ||
| Reporter: | wsmith (wsmith) | Labels: | |
| Date Opened: | 2004-04-05 15:00:31 | Date Closed: | 2004-09-25 02:52:15 |
| Priority: | Major | Regression? | No |
| Status: | Closed/Complete | Components: | Core/General |
| Versions: | Frequency of Occurrence | ||
| Related Issues: | |||
| Environment: | Attachments: | ||
| Description: | LibPri seems to die with the following error every 3 months for us (90k calls). It is hard to reproduce because it happens very seldom. The error is chan_zap.c:6250 pri_dchannel: Ring requested on channel 4 already in use on span 2. Hanging up on owner. and asterisk not longer takes inbound calls. We were running CVS as of 3 months ago and have just upgraded after experiencing the problem. | ||
| Comments: | By: Mark Spencer (markster) 2004-04-05 15:45:30 This says it dies every 3 months for you, how many times has it crashed 90k calls apart? By: wsmith (wsmith) 2004-04-05 15:55:47 Last Friday will be the 5th time it's done it. By: Mark Spencer (markster) 2004-04-05 16:57:15 Well, in order to debug it, I'm going to have to be able to login to the system. I doubt this has to do with libpri, but looks like there is some sortof locked channel on the Asterisk side of the equation. We recently (within the last week) had a fix for this kind of problem in CVS head (not sure if it's in -stable) so it might be worthwhile to update to CVS head. Next time it happens, I'll be happy to login and see what might have the channel stuck. If you can duplicate this within the next several days that would be great, otherwise I think we'll just have to close this until it becomes an issue again. It might be worthwhile to go through Digium tech support as soon as it happens so we can login and take a look. By: wsmith (wsmith) 2004-04-05 19:42:52 Sure. I would be glad to let you. However, 90k calls are kind of hard to generate over Zap channels. I will, reply as soon as it happens again. By: Mark Spencer (markster) 2004-04-07 10:01:09 This bug is believed fixed in CVS of libpri *and* Asterisk (depending on whether it was the deadlock bug or the double RELEASE bug) | ||