Hello!
I use FastAGI very frequently [meaning 30 channels IVR is made in it]
and sometimes I find, that there is a message like:
Jul 29 09:38:55 VERBOSE[896] logger.c: == Auto fallthrough, channel
'Local/1@route-out-eeae,2' status is 'CHANUNAVAIL'
Jul 29 09:38:55 VERBOSE[893] logger.c: > Channel
Local/1@route-out-eeae,1 was never answered.
Jul 29 09:38:55 VERBOSE[896] logger.c: -- Executing
DeadAGI("Local/1@route-out-eeae,2",
"agi://127.0.0.1/callhangup"
) in new stack
Jul 29 09:38:55 VERBOSE[590] logger.c: -- AGI Script
agi://127.0.0.1/callhangup completed, returning 0
Jul 29 09:38:55 WARNING[896] res_agi.c: Connect to
'agi://127.0.0.1/callhangup' failed: Bad file descriptor
Jul 29 09:38:55 VERBOSE[896] logger.c: == Spawn extension (route-out,
h, 1) exited non-zero on 'Local/1@route-ou
t-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Could anybody tell me what causes this 'Bad file descriptor'
message?>From the code I see, that it comes after the connection has been
established with FastAGI server, however I don't see anything on that.
This problem happens only very rarely [once/2days with continuous
30channels/8hours load].
What can cause that issue?
Did anybody think about using a unix socket for communicating asterisk
and the fastagi server? I know, we would lose the remote processing
feature, however we can save on IP stack when AGI requested handled locally.
Any idea how can I stabilize the FastAGI running? On the other side is a
python threading socketserver.
Thanks in advance,
Tamas
Hello!
I use FastAGI very frequently [meaning 30 channels IVR is made in it]
and sometimes I find, that there is a message like:
Jul 29 09:38:55 VERBOSE[896] logger.c: == Auto fallthrough, channel
'Local/1@route-out-eeae,2' status is 'CHANUNAVAIL'
Jul 29 09:38:55 VERBOSE[893] logger.c: > Channel
Local/1@route-out-eeae,1 was never answered.
Jul 29 09:38:55 VERBOSE[896] logger.c: -- Executing
DeadAGI("Local/1@route-out-eeae,2",
"agi://127.0.0.1/callhangup"
) in new stack
Jul 29 09:38:55 VERBOSE[590] logger.c: -- AGI Script
agi://127.0.0.1/callhangup completed, returning 0
Jul 29 09:38:55 WARNING[896] res_agi.c: Connect to
'agi://127.0.0.1/callhangup' failed: Bad file descriptor
Jul 29 09:38:55 VERBOSE[896] logger.c: == Spawn extension (route-out,
h, 1) exited non-zero on 'Local/1@route-ou
t-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Jul 29 09:38:55 DEBUG[893] channel.c: Avoiding deadlock for
'Local/1@route-out-eeae,2'
Could anybody tell me what causes this 'Bad file descriptor'
message?>From the code I see, that it comes after the connection has been
established with FastAGI server, however I don't see anything on that.
This problem happens only very rarely [once/2days with continuous
30channels/8hours load].
What can cause that issue?
Did anybody think about using a unix socket for communicating asterisk
and the fastagi server? I know, we would lose the remote processing
feature, however we can save on IP stack when AGI requested handled locally.
Any idea how can I stabilize the FastAGI running? On the other side is a
python threading socketserver.
Thanks in advance,
Tamas