MAX_AGI_CONNECT error

[spinto]

We have a multi-server install and are getting the following error on the second server:

[Nov 30 12:33:23] WARNING[23429]: res_agi.c:230 launch_netscript: FastAGI connection to 'agi://127.0.0.1:4577/call_log' timed out after MAX_AGI_CONNECT (2000) milliseconds.

Has anyone seen this before?

[mflorell]

might have something to do with DNS lookups and "localhost" perhaps

[Rumbles]

Hi,

I was trying to write slightly more thorough responses but the forum thinks I'm being spamy. I have also noticed this issue on a new machine in an existing cluster causing calls to drop. I believe this as users are reporting call issues at the same time as I see thousands of these messages go through the logs. I have updated the hosts file on the local machine to have it's own hostname against 127, but I am still seeing issues. Is there anything else I should try?

Thanks,

[Rumbles]

Ah, might have fixed it, I found I had made an entry on our internal DNS server for the hostname, which might have been causing my issues, I have removed it now, not seen a repeat, so I'll monitor it for a while. If I don't reply again please feel free to ignore my ramblings

[Rumbles]

I've been monitoring this problem over the past few days and I am still seeing the same issues, every now and then (but only when an agent is working on a machine/phone which connects via this new server) I see a spike in network traffic, followed by a slight lull. Sometimes when this happens the agents on the machine lose all audio, at which point we need to get them to log out and wait a few minutes before logging back in (if they log in too soon they still have no audio)

At the same time as this spike in traffic I see the same messages mentioned previously scroll through the logs hundreds or thousands of times (currently 30,000 entries in the logs on the machine and it's only been running a week)

Thanks for any help

[Rumbles]

I had to remove a lot of my post as the forum saw it as spamy, just to point out, the ptr and host records are now off the DNS server (if I do an nslookup for the hostname of my server it returns with nothing), and there is an entry against 127.0.0.1 in the hosts file for the hostname of the server. I'm not sure what else I should check, if there is anything that I should look at please let me know.

[Rumbles]

I have found the cause for this issue, it looks like there was a loop set up in the sip.conf, with another server with a different name (that used to use this IP) being referenced, on this server and others, so they must have been firing off requests to this server when it was out of use by IP address, and I guess it was then sending them on to the server name but then immediately receiving them again.

While this has solved most of the network spikes, the sound cutting out issue continues....