Polycom MWI & Dual Server


I have been battling with this on and off for years now and hope that someone can help as it’s finally reached the point of a user complaining.

We run a suite of Polycom IP500 handsets that are configured to register to a primary and a failover server.

Due to our config the primary does all the work and the secondary sits there on standby. Some users like VM to email but others like the old fashioned on the phone flashy light thing however this is causing a problem

When the phones register they check VM status, if a VM is left on the primary the MWI flashes and its fine however when the registration renews on the secondary it see’s no messages and the indicator stops. I can get it back by doing a reload on the primary which starts it flashing but if i do a reload on the secondary it stops.

I used to manually edit the voicemail.conf file but later found that if I empty this conf file completly it stops VM’s being left on the secondary which is exactly what I want. Messages could potentially get lost or out of sync if somehow left on the secondary so disabling it completely is the only way

I don’t want to go back to having to manually adding an entry per phone on the secondary somewhere. Ive seen some posts suggesting the below

/etc/asterisk/sip_custom_post.conf and add the following two lines per extension.

However this is something I need to remember to edit if the user changes or if anyone else wants the same.

Is there a way to stop asterisk notifying on a global basis on the secondary?

PS - I’ve tried everything under the sun regarding config changes with the polycoms but it keeps checking.

System is FreePBX

I added the two lines to sip_custom_post and it works perfect so I thought that I could add my entire number range (including extensions that were not yet created) and it would skip the ones that were not valid.

Category addition requested, but category ‘3000’ does not exist, line 3 of /etc/asterisk/sip_custom_post.conf

Unfortunately not. It skips the rest of the file once it comes across a number that does not exist which is a shame.

Is there a way to get it to skip or another way of doing this?

Another solution might be to rsync -a --delete /var/spool/asterisk/ with a cron job from the primary to the secondaryan added advantage that failing over registrationto the secondary finds an up to date state of the extensions.

If that command does what I think it does then you have two copies of the VM and that cron would need to run quite often to stay in sync.

I did try something similar with the backup module months ago but found that in a situation where the primary is offline then genuine voicemails were left on the secondary. The primary came back and everything switched back but then the vm’s were then lost or missed that were on the secondary. Phones would flash but they had no way of reaching the message as *97 would dial on the primary. It all gets a bit messy which is why I have completely disabled voicemail on the secondary. If a failover to the secondary occurs then people just have to do without voicemail until the primary is fixed.

If you failover that often maybe you should look at a more robust setup.

We’ve only fell over to the backup twice in 2 years and one of those was to take the primary off for critical upgrades

All it takes it one 5 minutes of downtime and a VM to get lost.

I like to think ahead and try and keep it as robust & seamless as possible, perhaps taking it too far but like to cover all bases.

Have you thought of deploying a real HA cluster. Personally I currently use clustered qemu servers under ProxMox with glusterfs file systems, milliseconds of downtime and no lost vmails, it’s always the same machine running, just on different hardware.