R81.10 - SmartConsole Central Upgrade with 3 member VSX cluster (R80.40 to R81.10)

[Gregory_Azratz]

Hi @caw001 

Awesome way to showcase the new way of upgrading machines from the comfort of the Smart Console.

Another thing I want to add that we are working on releasing new Blink packages for VSX upgrade.
This new package will allow our customers to perform the presented upgrade flow (R80.40 --> R81.10 + JHF) in a single action.
We are expecting to release this package in the upcoming months 

Gregory

[caw001]

@Gregory_Azratz - In my video, I only demo'ed 1 VSX cluster.  What is the maximum number of clusters that customers can perform upgrades on simultaneously?

[Gregory_Azratz]

The specific number is written in the Admin guide of the MGMT as it might change from version to version - link

For R81.10 its

"You can select up to 30 Security Gateways and Cluster Members, but installation can take place only on 10 targets at the same time. The Management Server places each target above the 10th in a queue. Each time an installation completes on one of the targets, the Management Server installs it on the next target in the queue."

[KostasGR]

Hello @caw001

Nice video! Can you please create another video following  Multi-Version Cluster Upgrade Procedure - VSX Mode (VSLS)?

BR,
Kostas

[aheilmaier]

think that video was a 3 node cluster in vsls mode.

[Magnus-Holmberg]

Nice to see that its getting some progress on it making it easier for everyone.

However just in the process of upgrading 15 VSX clusters with 2 - 4 members in VSLS R80.30 3.10 to R81.10
And sadly its getting active / active on a few times meaning full production outage, so we stopped with that.
(not related to sk174510 as R80.30 3.10 should use the same as R80.40.)

But i will give it a go again in a few months 🙂
May be due to some priv hotfixes, so we decided to go for clean install and MVC worked much better.
Clean install also saved about 1-2hours of the maintenance window due to boot time with large VSX clusters.
Also went for R80.40 due to some bugs first fixed in ongoing hfa on R81.10,

Still have some requests here.
- Possibility to keep 2 of 3 members active (more or less pick how many members upgrade same time)
As 1 may not manage the complete load.

- Pick the actual failover, so it fit better with maintenance windows.
It look very seamless here but put 30->50 VS in the cluster and you see the time increase by alot.
Its much much faster after running R81.10 on the MGMT server, before it was a nightmare.

- Pause the upgrade, related to above to provide time for testing.
- Install the JFA before continue to use the node, (see that this was covered below, Thanks!)

Regards,
Magnus