Hot spare and massive volume of call recordings

So, I am in the process of building a hot spare setup. The client will be recording all calls and keep them for a year.
I have 6TB of space (3 2TB diks) in each server to store everything.

From some tests I run, the backup that I have configured as hot swap doesn’t copy the recording files.
Well I prefer it this way, making a backup of the configuration and moving with scp all files to the spare server.
But how you would tackle the moving of the files between the 3 disks and keeping cdr reporting happy to be able to retrieve the file through the gui.
Any recommendations are welcome.

if you have the space, set up an NFS Mount, and map that to your /var/spool/asterisk folder on both servers.
that way, all recordings are only in 1 location.

we do the same with our endpoint config files.

From previous installations and documentantion mounting a network share as the destination folder for MixMonitor is a bad idea that might cause recordings with unsync parts.
I would prefer first make the recordings in a local folder and then move it to a network share.

i would disagree with that, based on how NFS Mounts have worked for us in the past, but fair.

Ha, so you have some experience with recordings, you mentioned config files which is just a small file that is waiting to be served.
So you had success with this kind of solution. I though doing something like that but I don’t have a third storage server, just these two.

You don’t, period. You properly use RAID or LVM or similar to create a single 6TB store from the 3 disks, and then you properly mount that. as the recording location.

1 Like

our recordings and endpoint configs both use this.

Worst case, the current active calls don’t get processed 100% during a server failure.
which … the call will die when the server does anyway.

best case, all of our recordings transfer to the spare.

There are a couple of things to add to /etc/asterisk/asterisk.conf to alleviate stale nfs or otherwise disappearing network shares

cache_record_files => yes
record_cache_dir => /private/tmp

where record_cache-dir is local and well sized.

Look into glusterfs if you are limited to two machines and want the share to be bi-directional (replicated in glusterfs-speak). I used zfs as the underlying drive format.

1 Like

Well I though using RAID but to use all available space I must configure it as RAID 0 which I think it is not a good idea. On the other hand LVM seems the solution to go. If I loose one of the disks what happens to the LVM. I have never used a volume spanning three physical disks.

@Dicko I have never used glusterfs, can you share any link that you have used before.

Would be a good place to start. For zfs arrays of disks and the why’s

Is a good overview.

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.