Migration in failed situation without HA ?

[crazy_monkey]

Hey guys, i was reading about proxmox HA, and i found out that for truly HA you need some decent hardware, like speeds 2.5G+ , but in my situation i would just like to have a backup solution that spins up in case of a failed hardware of main server for just opnsense. Basically i have opnsense in a vm in proxmox, its running fine , but if something happens to this machine i will be in a bad situation losing internet in my home, and would need to spend so much time bring back.

I would like to have a second machine that spins up in case of failure of the first, it dont need to be super fast like HA, would be ok to have like 15 min of downtime, and i only need this to opnsense vm

Whats paths do you guys recommend i look to solve this situation without spending extra money with expensive hadware for HA ?

thanks in advance !

 

[alexskysilk]

Basically i have opnsense in a vm in proxmox, its running fine , but if something happens to this machine i will be in a bad situation losing internet in my home, and would need to spend so much time bring back.

Almost never a good way to go. Your router should be independent of everything unless you are sophisticated enough to set up your own vrrp pair. in a home environment... the standalone router will be more dependable

I would like to have a second machine that spins up in case of failure of the first,

If you mean automatically- you're back to a cluster. if manually- the downtime will be the duration between being downed and when you get around to fixing the situation either by bringing up the spare or fixing the main.

 

[crazy_monkey]

If you mean automatically- you're back to a cluster. if manually- the downtime will be the duration between being downed and when you get around to fixing the situation either by bringing up the spare or fixing the main.

Is there a cluster without HA, that is automatic BUT normal hardware could handle (like 1gbps ethernet ports) ? i dont need fast transition, could take some time if necessary but being automatic is a must

I haven't defined the solution yet, just wondering the pros and cons

 

[alexskysilk]

Is there a cluster without HA,

Yes.

But you want the opposite- you WANT HA whether you deploy a cluster or not.

You can (and probably should) be thinking of clustering your ROUTERS since thats the only thing you seem to want for high availability anyway; and yes, that can be done via the aforementioned vrrp pair. you can find instructions in the opnsense manual. or on the interwebs. The routers can be vms but they would need to be on edit- seperate hardware for the solution to make sense.

 

[crazy_monkey]

Yes.

But you want the opposite- you WANT HA whether you deploy a cluster or not.

You can (and probably should) be thinking of clustering your ROUTERS since thats the only thing you seem to want for high availability anyway; and yes, that can be done via the aforementioned vrrp pair. you can find instructions in the opnsense manual. or on the interwebs. The routers can be vms but they would need to be on edit- seperate hardware for the solution to make sense.

sorry, i dont get when you say ROUTERS, i just have OPNsense vm regarding routing/firewall/vlans like the backbone of the network , and then a ethernet cable send the vlans via trunk to physical openwrt routers (tp-links)

edit: i think i got it , when you mean routers , you mean opnsense in each hardware, right ?

Also i found a possible solution : https://forum.opnsense.org/index.php?topic=44697.0 , the guy uses zfs replication , look less complicated than vrrp pair , would say this could be reliable with zfs replication ?

 

[alexskysilk]

and then a ethernet cable send the vlans via trunk to physical openwrt routers (tp-links)

umm... I dont understand your network. why do you need cascading routers? EITHER your WRT or Opnsense can and should manage your entire network.

Also i found a possible solution : https://forum.opnsense.org/index.php?topic=44697.0 , the guy uses zfs replication , look less complicated than vrrp pair , would say this could be reliable with zfs replication ?

It could work, but this describes a proxmox cluster which I thought you wanted to avoid.

Seriously, why are you looking for so much convolution? just use your WRT router and call it a day- and since you said "routers" you can set them up for HA. see https://openwrt.org/docs/guide-user/network/high-availability