FreePBX latest distro on HP DL 380 G8 server

Hi

Does freepbx 64bit latest ver can be installed on HP DL380 G8 server?

Thanks!
Slamis

Yes, it can. Are you hosting about 50,000 phones?

Make sure you install the HP ASM and RAID CLI utilities for Linux along with the enhanced HP MIB

Hi

Thanks!

I have about 500 phones but many concurrent calls (about 200-400)
I need also many storage for the call recordings…

Id the HP DL380 is too powerful?
Any other recommendations?

IMHO you can never have a server that is too powerful. :wink:

I think a better title for this post would be “I have 500 extensions and 200-400 concurrent calls. What kind of hardware, network design and software will I need?”

I have never deployed an Asterisk/FreePBX based system of this size so I don’t really have any direct experience. I can assure you that something of this size requires a bit more planing and design than just plopping a FreePBX server in the mix.

Regarding the subject - you are right :slight_smile:

What about HA module?
Servers need to be the same hardware?
The HA is good redundancy solution?
Im also using Queuemetrics as my Call center SW will it work with HA?

Any other cloning solutions? (I mean to copy all server data/cfg to another standby server?)

Thanks!

Any DRBD based HA system is limited in as you can only have two working nodes, you could look into iscsi or glusterfs as the holder of the “state” of your machines, both allow multiple failover redundancy, your Hardware is ideal to support either, glusterfs being much easier to setup and maintain.

If you have the ILO hardware then your cluster-manager when set up to properly fence your machines becomes very stress free.

Your CPU supports both VT-d and VT-x so if you have any dahdi hardware cards you can “pass them through” to any underlying “virtual” machines.

One very easy solution might be to install ProxMox on two or three machines and set up any number of FreePBX/Asteri under the cluster either manually or from generic distribution , no license fee or special considerations needed, the machines will transition seamlessly on hardware failure within a second possibly without loosing your current SIP connections, hardware a few seconds longer if there are any TDM links.

With the load you anticipate, it is probably wise to investigate and increase as necessary any ulimits as asterisk is quite greedy with it’s usage per open channel involved.

For those without VT-d you can still proxy your dahdi type hardware with dahdi’s dynamic_eth driver, and dacs the channels over to the virtual machines.

(end result of the above, a fully GUI managed expandable redundant system with backups and snapshots. Add a SIP proxy machine or two and a whole farm of separate PBX’s all working together without problems)

(and No as to identical hardware, you can successfully cluster hardware from any manufacturer Dell,P,IBM etc. provided they have VT-x or equivalent, and ipmi support, Gbit or higher network infrastructure is highly recommended :slight_smile: )

Of course we are biased but FreePBX HA works awesome and lots of our RPMs and things in FreePBX have been tweaked to make sure when you update FreePBX or RPMs they do not stomp on your HA setup which will can and will happen in other HA setups.

Without doubt drbd is awesome, I have used it for five years in my FreePBX soultions, it’s just that glusterfs is more awesome and way easier to setup and truly scalable unlike the older master/slave DRBD somutions, you should try it sometime, it will only take about two hours to do. You can even do it under ProxMox itself, and that is REALLY awesome.

Virtualizing the whole machine and not just the state makes sense in so many ways, only one machine to update, only one to backup only one to restore, the machine is a single instance of a standard FreePBX, perhaps even “The” FreePBX distro :slight_smile: the machine maintains one mac address so any licenses like g729 need only be bought once. Clones backups and snapshot are migratable.

For example it takes a few seconds to spawn a new FreePBX instance using qcow2 against a glusterfs backing disk peviosly “templated”, it will take up minimal disk space until data is added.

Come on you adventurers out there, give it a bash and report your experiences, nothing ventured, nothing gained.

(the cloned instances of course will have a different mac address, so get your wallet out when necessary and carefully guard your original machine before converting it to a template unless you want to update and re-template your hard work)

(and it’s not cheating any license holders, because it is only an instance of a standard non HA machine, only one copy can run at once, well, two could but if they did you would break things, so don’t try that at home folks :wink: )