Asterisk 1.2.15 w/ zaptel 1.2.13 or 14 is suddenly crashing, the only major change was kernel update to 2.6.9-42.0.10.EL from 2.6.9-42.0.8.EL (CentOS) No config changes.
I always rebuild libpri 1.2.4 zaptel and asterisk with kernel updates.
I have added PROC=pentium4 to the Asterisk Makefile for some time. Otherwise it is pretty stock, mg2 echo can.
I was trying to track it down, and could easily replicate with a parked call. asterisk outputs the following.
channel.c:929 ast_channel_free: PBX may not have been terminated properly on ‘Zap/4-1’
== Spawn extension (macro-stdexten, s-NOANSWER, 1) exited non-zero on ‘Hn**’ in macro ‘stdexten’
== Spawn extension (macro-stdexten, s-NOANSWER, 1) exited non-zero on 'â–’::'
Mar 2 11:04:51 WARNING[26377]: chan_zap.c:2462 zt_hangup: Asked to hangup channel not connected
Mar 2 11:04:51 WARNING[26377]: channel.c:897 ast_channel_free: Unable to find channel in list
*** glibc detected *** double free or corruption (!prev): 0x089f4af8 ***
No other changes, been stable forever, otherwise. Am rebuilding on previous kernel now, but suggestions would be welcomed.
I will follow up if I see any new details. I did notice zttranscode is loading once asterisk is started, which I hadn’t noticed in the past.
I haven’t used parked calls in a while, after rebuilding with previous kernel I see the following. A call came in in the middle of this, and we crashed again. I will avoid parked calls for now… the moh ends up in voicemail… zap/4-1 is fxo, zap /1-1 and 2-1 are fxs
Mar 2 11:29:14 WARNING[10928]: chan_zap.c:4660 zt_read: Thread -1209812048 Blocking ‘Zap/4-1’, already blocked by thread -1209812048 in procedure ast_waitfor_nandfds
– Started music on hold, class ‘default’, on channel 'Zap/4-1’
Mar 2 11:29:15 WARNING[10928]: codec_gsm.c:165 gsmtolin_framein: Huh? A GSM frame that isn’t a multiple of 33 or 65 bytes long from (null) (320)?
– Stopped music on hold on Zap/4-1
– Registered extension context ‘park-dial’
– Added extension ‘Zap/2’ priority 1 to park-dial
== Timeout for Zap/4-1 parked on 701. Returning to park-dial,Zap/2,1
Mar 2 11:30:40 WARNING[10966]: pbx.c:2250 __ast_pbx_run: Zap/4-1 already has PBX structure??
– Executing Dial(“Zap/4-1”, “Zap/2||t”) in new stack
– Called 2
Mar 2 11:30:40 WARNING[10966]: channel.c:1604 ast_waitfor_nandfds: Thread -1212781648 Blocking ‘Zap/4-1’, already blocked by thread -1212515408 in procedure ast_waitfor_nandfds
– Zap/2-1 is ringing
– Zap/2-1 is ringing
Mar 2 11:30:44 WARNING[10941]: app.c:644 ast_play_and_record_full: No audio available on Zap/4-1??
– User hung up
== Spawn extension (park-dial, Zap/2, 1) exited non-zero on ‘Zap/4-1’ in macro ‘stdexten’
== Spawn extension (park-dial, Zap/2, 1) exited non-zero on ‘Zap/4-1’
– Hungup ‘Zap/4-1’
– Hungup ‘Zap/2-1’
== Spawn extension (park-dial, Zap/2, 1) exited non-zero on 'â–’::'
Mar 2 11:30:44 WARNING[10966]: chan_zap.c:2462 zt_hangup: Asked to hangup channel not connected
Mar 2 11:30:44 WARNING[10966]: channel.c:897 ast_channel_free: Unable to find channel in list
*** glibc detected *** double free or corruption (!prev): 0x08fcfaf8 ***
Aborted
I’ll keep toying with it, but suggestions are welcome.
I moved up to 1.2.16 last night. I think it is hardware because other processes are having a bad time - it’s not just asterisk. Power looks good, but there is something bad, I will try to swap out memory this afternoon.