Minemeld High Availability

[rchilukuri]

MineMeld needs to be deployed in a high availability model, so if one goes down firewall will not start blocking allowed traffic from the rule base. What is the recommendation here?

[lmori]

Current suggestion for HA is having 2 MM instances running behind a load balancer. Config sync should be still done manually.

Luigi

[lukjdye]

If we don't want to put ti behind a LB, what about having two MM instances, and having PAN pulling both feeds seperatly then incorporating both fo the feeds into the rule base? Do you see any issues with that?

[lmori]

Hi @lukjdye,

Great Idea ! I haven't tested this yet, but I think it should work.

Luigi

[irt-unimi]

Using two MM instances and two EDLs for the same IP lists adds overhead to the firewall and could hit the EDL IP number limit for your hardware.

If an EDL source URL is unreachable, PAN OS just keeps using the old EDL values, so I would not be overly concerned if you can endure it 24-48 hours.

I would rather like to be warned when MM goes down, so I suggest configuring something like Nagios to ping and check https.

And I would like to be notified if a Miner fails updating a list N times: right now it seems the alert is only shown in the web UI.

[lmori]

Good point @irt-unimi. The double EDL solution would be viable only for small EDLs, like O365. 

[mmathis]

Is this still the reccomendation?

[lmori]

Yes, but we are planning for something more advanced for the next major.

[mmathis]

Thanks! any rough ETA on the next major release?

[nogueirad]

Just chekcing in to see if you had a target date for this.

Thanks!