frequency wrote:Hello,
Ofcourse I searched before posting. I am basically having issue on multiple servers. basically on both multi and single servers. 5 to 15% calls are going to LRERR status. Carrier is showing call as dialed for about 8 seconds then it drops.
I know this happens due to firewall issue or a bad carrier or bad connectivity whatsoever but that is not the case. Even with the firewall and iptables completely off, it is not the case. I even have a server which is suffering from this issue and is in the same dc but most of them are.
Multi servers are working with similar specs, same build. MTR Seems fine with zero losses with Load average hovering around 1.00 to 2.50 when spiked up.
Server specs: E3-1271v3, 32g ram, 480gb ssd, Asterisk 1.8.23-vici, VERSION: 2.12-564a BUILD: 160809-1339, No firewall, scratch installed.
1) Please show a link to your installation instructions. If this is an installation issue, obviously this can help others with the same installation method find these results, and perhaps will result in someone altering those instructions to resolve future issues.
2) When "5% to 15%" calls are affected, it is often easy to start trying to resolve general items to solve this general problem. But I strongly suggest you pick ONE call and troubleshoot it by checking logs (which may require turning all logging methods on temporarily) to see what the reason for the call termination actualy is. In code (while the pretty 'non-vicidial' chart is nice to look at, it doesn't contain any termination codes ... just the 'bye' which is what happens after the code is received/generated).
3) You have tossed some vague "we have lots of systems and some don't work" information out there, but you left off something that would be useful: Since you are experiencing this on both single and multi-server installs ...are you then experiencing it on ALL systems? Or only some? All single? all Multi-server? Or only some of them? What do the servers experiencing this have in common that are not shared with the servers not experiencing this issue?
4) "MTR Seems fine" ... this, combined with the pretty chart above the statement are the hallmarks of "this guy knows enough to be dangerous". What I mean by this is that you are likely an IT professional with a lot of networking experience and are experiencing a networking issue that other (non-IT-professional) call centers do NOT have. You undoubtedly have a very cool network configuration. Powerful. Flexible. Different from everyone else. And you're experiencing a networking problem that defies (your) logic.
5) You did not mention whether these servers have Public and/or Private IPs. No mention of NAT vs No NAT.
6) Just for clarity: Lots of servers (single and multiple-server systems), but only one vicidial/asterisk version listed. All your servers have the same versions of asterisk and Vicidial?
My best guess: based on your description, you are very likely the cause of the problem by having a cool network configuration. LOL. Pick one of the systems experiencing the issue, put it on a public IP with a generic ($35 Walmart) router or direct to a public IP with NO router at all. If the problem goes away, you may want to rethink your cool networking where Vicidial is involved.
Vicidial is not "just an average PBX". It is capable of generating thousands of calls per minute and using up router resources in ways that other systems do not. MySQL runs out of links/connections. Asterisk, Apache and MySQL use up massive quantities of ports. Somewhere in error logs is your answer.
