Let me start with what is going on, and apologizing with how long this is, I just want to provide all I can upfront and be detailed in the issues and attempted fixes. We just moved buildings and decided to upgrade to the latest FreePBX since the version installed was from 2009. We set up a Dell Server with 12GB of RAM, running a minimal CentOS 7.4.1708 install. Software is FreePBX 188.8.131.52, Asterisk 15.2.0, and Dahdi 2.11.1+2.11.1. We are using a Digium 8 FXO AEX800 PCI-E Card with WCTDM24XXP module and VPMADT032 Hardware Echo Cancellation. We are running one DAHDI line into a hunt group of four total numbers and PJSIP extensions. When the phone goes down, the PJSIP lines still function with the ability to call extensions, but the incoming DAHDI line does not work until we run “fwconsole restart”. I have made it a point to call in 4-5 times a day, to check the status of the phones, which is not a solution for moving forward.
There are multiple issues that have presented itself since this install has gone live the past month. Almost occurs daily or every other day. Sometimes when people call in they are unable to hear the IVR. Other times people call in and it just rings and rings and rings without stopping and do not reach the IVR or default route set up by the IVR. Sometimes, people can call in, and the IVR answers, but does not accept any of the inputs on the IVR but if let alone, will reach the default route. The last of the reoccurring issues is when people call in, and our side answers, the calling in side cannot hear the answering side and it appears to be one-way sound.
We have been troubleshooting and scouring many of the WikiPages of help and many more help pages and feel like we have tried everything. We are hoping the community will be able to assist, before we attempt a rollback or reaching out to Digium.
The following have been error messages observed in dmesg which seem to be occurring prior to finding out the phones are down.
- Missed interrupt. Increasing latency to 5 ms in order to compensate.
- Unable to disable sw companding on echo cancellation channel 0 (reason 4)
- Unable to set SW Companding on channel 0 (reason 4)
Some of the things we have attempted to do to resolve the issue include, but are definitely not limited to:
- fwconsole restart fixes issues most of the time
- Changing from Software Echo Cancellation to Hardware
- Upgraded Server’s BIOS
- Latest Dahdi firmware
- Set relaxdtmf to ‘yes’
- Used fxotune to attempt to configure
- Disabled the system frame buffer by adding nomodeset to the kernel boot line per the following knowledge base article: https://support.digium.com/community/s/article/How-to-disable-the-Linux-frame-buffer-if-it-s-causing-problems
- Installed acpid.
- Made sure irqbalance was installed
- Installed OS updates and rebuilt the kernel module
- Ran the fxotune command again with a silence timeout of 2 seconds
Issuing systemctl restart dahdi did not fix the issue. Only fwconsole restart did.
The only way we have been able to fully restore phone functionality would be to run “fwconsole restart”. This has also occasionally caused the FXO Port groups and Context to need to be reconfigured under DAHDI Config.
Any help or suggestions would be greatly appreciated!