Migration from Classic LB to External Managed LB - 502

Google Cloud Compute Infrastructure
1

Hi,

I tried to apply the migration from the classic load balancer to the newer external managed load balancer.

The classic has been working well for a couple of years and is connected to numerous Cloud Run instances. I went through the migration process and tried forwarding % of the traffic via the “migration” process. But any traffic that goes through the new load balancer returns 502.

There’s no explanation - only that it’s a temporary issue and I should try again in 30 seconds. Load balancer logs just show the 502, with no useful details. No logs in Cloud Run.

Has anyone been successful in running the migration process? Any suggestions?

In addition, the network analyzer is saying that the Cloud Run instance can communicate with the load balancer.

Reverting to classic fixes the issue.

At this stage, I’m leaning towards a GCP issue, but without paying for a support tier, there’s no way to reach anyone at Google Cloud.

Extremely frustrating situation.

Thanks!

2

So it’s a bug for sure.
The new LB starts working the moment I tell google cloud to complete the migration.
In other words, the migration is broken. There’s no migration…

It takes 6 minutes to kill the classic load balancer and enable the new one. Which means it causes around 6 minutes of downtime.

I just applied the changes to my production. It took down all my services for roughly 5 minutes (I migrated everything at the same time). But now it’s working. Risky for sure, but I wanted it to be done with it.