Firmware upgrade HA cluster causes downtime?

[denji]

Hmm…I have this issue with the hardware devices and I just follow the guide to create a separate Mgmt interface for each cluster member and then I can do a zero downtime firmware upgrade.

However, you can’t add any Vnics to the azure Fortigate appliance.

Does anyone know of a workaround so one can do firmware updates with zero downtime when running the HA azure appliance?

[gfleming]

Uninterrupted Upgrade is possible with standard HA config. I don't believe you need additional dedicate MGMT NICs for this to work:

https://docs.fortinet.com/document/fortigate/7.2.3/administration-guide/247944/upgrading-fortigates-...

[sjoshi]

Dear denji,

Thank you for posting to the Fortinet Community Forum.

Problem Description:-

Firmware upgrade HA cluster 

Please follow the link below which explains how the FGT upgrade works on HA cluster.

Yes with zero downtime you can upgrade the HA cluster but there will be few packet drops when it switches between Master and Slave

https://docs.fortinet.com/document/fortigate/6.2.12/cookbook/247944/upgrading-fortigates-in-an-ha-cl...

Let us know if this helps.

Thanks

[aahmadzada]

Downtime also can be due to upstream/downstream switches failing to properly update the CAM table when the failover occurs.
The same is applicable to the arp entry update on the upstream and downstream L3 devices

Ahmad

[Shilpa1]

Hello,
With the FortiGate appliance in Azure,  zero downtime firmware upgrade while running the HA cluster, needs a different approach, since you cannot add any VNICs to the Azure FortiGate appliance.

One possible workaround is a rolling upgrade approach, where you upgrade one cluster member at a time, while the other cluster member is active and processing traffic. This approach ensures that traffic is always processed by at least one cluster member, providing zero downtime for your applications.

Here are the steps to perform a rolling upgrade:

1. Take a backup of the configuration on the active cluster member.

2. Disable HA sync on the active cluster member to avoid the configuration being pushed to the passive cluster member during the upgrade.

3. Upgrade the firmware on the passive cluster member.

4. After the upgrade is complete, verify that the passive cluster member is running the upgraded firmware and that it is syncing the configuration from the active cluster member.

5. Make the upgraded passive cluster member the active cluster member.

6. Repeat the same process for the other cluster member.

This approach ensures that only one cluster member is down at a time, and there is always one active cluster member to handle traffic. However, it is important to note that during the firmware upgrade process, there may be a brief interruption in traffic as the active cluster member fails over to the passive cluster member.

Another option would be to use a load balancer in front of the FortiGate cluster to distribute traffic between the two cluster members. This would allow you to perform a zero downtime firmware upgrade by upgrading one cluster member at a time, without any interruption in traffic. However, this would require additional configuration and infrastructure, and it may not be feasible depending on your specific requirements and environment.

Regards,
Shilpa