Hi,

I'm testing cap devices managed from capsman, and I'd like they would be as independent from capsman as possible. My idea is that once caps have their configuration, they work independently from capsman, and to use caps manager when I want to change any configuration. With that way of working, I'm trying to avoid that capsman crashes produces the wireless network falling apart. Unifi controller uses a similar procedure. Would it be that possible?

Regards,

I agree. In scenarios where CAPsMAN is not involved heavily with station traffic [eg local forwarding, WPA2-Personal], CAPs should not freak out because they can't see CAPsMAN.

Consider also a scenario where CAPsMAN is centralised and maybe WAN link is congested. There should be a way to tune how long CAP takes before it declares CAPsMAN is dead and decides to withdraw services.

Maybe this sort of "survivability" could be automated by nominating a CAP to be a backup CAPsMAN without having to manually program it.

right now... you get a couple seconds before the CAP considers itself disconnected. Then the config disappears. CAP drops back to waiting for configuration.

I have considered some sort of back up scenario. Where the AP has the config in cap and it is set as the back up.

I have considered some sort of back up scenario. Where the AP has the config in cap and it is set as the back up.

What we need is CAPsMANMAN to sync configs between main CAPsMAN and backup CAPsMAN. Maybe in v7.

I second this. I have hotels that have hAP AClite connected to a 921GS-5HPacD r2 on the 5ghz. Capsman is unreliable in this scenario. 5ghz ->as a backhaul to the 5ghz hAP Capsman on the 2.4

Bill

This persistent mode would be good ONLY for CAPSMAN client configurations for local forwarding enabled and maybe be able to select an expiry period that would cope with your dropouts. I think it'd be better though to nail down why your CAPSMAN link keeps dropping out. I have found in the past that replacing crummy old legacy *Link brand switches on a site that had terrible CAPSMAN dropouts solved the whole problem I was having.

Just a thought which I haven't tried - did using CAPSMAN in IP mode rather than the default broadcast domain mode help improve reliability?