Configuration replication? Clustering?

I saw a line in the change log

net - apply slave config only if master config has been changed

What does this mean? Is it the beginnings of support for configuration replication between a master and a slave router?

would be cool. curious myself too.
would make sense if its rely on one of fully-distributed DB, like mnesia or alikes, ie things “without point of failure” and perfect scalability(aside failover and reliability benefits).

No, I believe this has to do with setting ports as Master / Slave on a switch chip.

boo

Agreed. Maybe someday they will be able to do High-Availability. That is the one feature that I miss from SonicWALL. But once Dell bought SonicWALL, I no longer trusted them.

Have guys guys tried to roll your own? I’m roughly playing an automated backup of one router to another one and then use a VRRP directly connected interface to detect and takeover/giveup all the interfaces (except for the directly connected VRRP one). It roughly seems to work OK, I lose some private keys (like ssh) and I haven’t sorted out how to determine when to ‘send’ the backup but I am roughly able to switch between master/slave. The only downside is that it isn’t pushing binary changes or anything, it reboots the second router to get it to takeover.

Curious if anyone has tried to pursue this and have any insight into the idea.

PS: I have an on startup script correcting mac addresses and switching to the proper A/B information. You might have noticed this as a glaring problem :slight_smile:

Bummer… I really would like to see a capability to mirror certain parts of a configuration from a master to a number of slave routers…
(would also be OK when the master had to be e.g. a tftp or http server)

In my case, I would like to mirror /ip firewall address-list for the items of specified list= items and
/ip firewall filter for the items of a specified chain= (that can then be called from forward and input chains)

+1 for this from me, too. It would be VERY useful for hostile but key sites, such as mountaintops. If the end devices only have one port though, how can this be done, without having a single point of failure (which can be, say, a middle switch)?

I have released this for testing:
https://github.com/svlsResearch/ha-mikrotik