I was wondering if anyone has setup a FreePBX distro installation with the recordings being on a remote NFS Share. Are there any implications? Is it a good idea to do that?
Our system has around 60-70 concurent calls so there will be alot of NFS reading / writting.
We had a case where the FreePBX HA thought asterisk was down and tried to restart it resulting in asterisk crashing and node b to taking over the cluster. I think it was our NFS share was that made that problem and stalled asterisk to the point it did not respond to FreePBX HA services. Now we save our recording locally and move them when needed. The FreePBX server and the “recordings” server are linked with a Cisco Catalyst 10/100 switch (old model - can’t remember which)
@dicko
So in our situation, if we want to use NFS again you suggest trying cache_record_files option? Both node A and node B have a NFS share to the “recordings” server.
So, If I understood correctly, this parameter will save the temporarily recordings until it will gain access to the NFS share?
It really sounds as a good backup for NFS networking problems.
That is the intention , but a flapping nfs mount will sooner or later leave you with that stale file message in which case a reboot is the only safe and practical recourse.
Well, FreePBX as stated at the documentation uses DRDB:
Free PBX High Availability (HA) is a commercially developed High Availability solution that has reworked the FreePBX platform to integrate DRDB, Cluster Manager, and Pacemaker
I understand, but I don’t use FreePBX HA , I moved away from drbd some years ago.
For efficiency, I suggest that if you go “native” with your syncing, then you use rsync for that, again for efficiency and robustness. You would in fact then be using a delayed form of sshfs.