gheto redundancy

[s2svoip]

I am looking at ways to add some redundancy to my setup, I don't have a membership so have not configured BDR and such but what I have managed to do is spin up a second server in AWS that has an auto restore from the live server every 24hrs. My thinking behind this is if I have a problem with the primary server I can just re-assign the public IP to the backup server and it will instantly take over - this seems like a better solution than DNS failover as that has propagation delay.

My question is, how could I utilize the 2nd sip server entry that most handsets support, so they register the same line but on another server. if my domain is pbx1.domain.com and that points to the public IP and then in fusion I have that domain, if I have a mirror of that same server on another public IP & dns, say pbx2.domain.com this wont tie up with the domain in fusion anymore

is this even possible? I know there are much better methods to do this, just working within the confines of my knowledge

 

[gmaruzz]

as for ghetto stuff (I like ghetto stuff), why don't you script an aws code that each night stops the VM, get a snapshot, start the VM. (ditto for possible volumes).

then you ship snapshots and volumes to another aws zone.

In case of disaster, or aws zone is down, an aws script will launch snapshot (and volumes) in that other aws zone

don't forget to delete old snapshots/volumes, you maybe want to keep three days, and one month ago (for a total of 4 snapshots/volumes)

all this can be done with some dozen perl lines (with aws perl api), and can be in some way activated by a check on aws services (or route53) again via perl or other scripting api to aws

If you're careful using low cost things (S3?) this will cost you almost nothing, and give complete disaster recovery as a bonus

 

[gmaruzz]

obviously, in case of activation of other zone, you first start snapshot (and volume), then you change route53 aws dns for the domain, again via aws api (perl or whatever)

 

[s2svoip]

thanks for the tips. yes if my knowledge of AWS was a little better, I think automation and redundancy using this is a better route for sure. a wise man once told me the best system is the one you know.

I actually made quite a bit of progress with my setup. My primary fusion deployment backs up daily to S3, I then have an install in azure that pulls that backup and restores it daily, I then use the 2nd sip server option on handsets for failover. this combined with my sip trunk provider having an endpoint priority setting I have full live redundancy across 2 cloud providers inbound and outbound - tested it and it's pretty much seamless.

I am sure there are better setups but this is what I have setup and understand. a live db mirror might be better, but then if a config error is made it replicates. I have a 24hour restore delay so that could assist with that.

the only thing left I want to do is move my primary domain to route 53 so I can move the dns in event of failure, I have a few POTStoVOIP box's that don't support 2nd sip server so think it would help with that

 

[gmaruzz]

honest, what i delineated is way easier (and probably less costly)

maybe you want to look into that, and do a little bit perl aws, or ansible aws experiments

anyway, and in any case, always keep *at least* the last three days *and* one month ago

so you can find info and stuff that were wrong since longtime...