Azure HA Failover not working

[RyanJohnstone1144]

Hello,

We have a pair of VM300 PAs in Azure set up in Active-Passive.  They are running 9.0.7 code with VM Series plug in 1.0.8. 

There was an issue in Azure on 19/10/20 which caused a failover and recovery (we use pre-emption).  Post this issue the PAs were up and running but not passing traffic.  we found that the secondary IP addresses (i.e. floating IPs) had been moved to the Azure VM for the Passive firewall (PA2).  Hence no traffic flowing as this firewall was passive.  We failed from the Active (PA1) over to this Firewall (PA2) and some traffic started to flow but everything was incredibly slow.  we tried restarting VMs, failing back over, etc but nothing would change the state of the secondary IP addresses, they were locked to PA2.

Eventually we completely powered down the VM which was running PA1 and things started to run ok again.  we then configured PA2 to always be Active and powered PA1 back up.  PA1 came back up, re-established HA and things were running fine, PA2 Acitve, PA1 Passive.  We then suspended PA2 to trigger failover and again we had issues with secondary addresses.  The secondary addresses on the untrust VM interface floated over to PA1 correctly, but the secondary address on Trust VM interface disappeared completely from both PA1 and PA2.  No failovers, restarts etc recovered this address.

We powered down PA2 and had to manually re-create the secondary address on Trust on PA1 to restore service.  this is the state we are now in.

can someone please assist / recommend next steps?  Failover it seems is broken.

thanks

Ryan

[RyanJohnstone1144]

HI there, no solution no.  am pending some assistance from PA TAC once i can arrange an outage.  We have tried multiple code versions and plug ins but problem remains.  i will update once i have an update, hopefully in the coming month.

[RyanJohnstone1144]

still no "solution" to this but carried out a controlled failover last week and it worked, first time it has worked as it should have done since the PAs were stood up about 2 years ago.  FYI the code we are on is 9.0.13 Azure plugin 1.0.13.

have some upgrades coming up over next few months to get to 9.1 so be interesting to see if any issues occur then.  otherwise can only assume something was fixed in the background Azure end or maybe i just got lucky...

[JimMcGrady]

I have always found this failover mechanism within Azure to be impractically slow. I look forward to the day when it improves.