2020-11-02 12:42 PM UTC: Migration is done.
Unfortunenalty the new nodes didn’t start properly due to a misconfiguration. In Squidex v5 a new setting was introduced to not accept other host names except the configured base URL. This setting was set to true as a mistake. Therefore the readiness probes and liveness probes were not able to read the health endpoints, because they make a request to an IP, not to the public host. It took a while to figure this out, which is unfortunate, because the original migration was done within 15 minutes.
2020-11-02 12:43 PM UTC: Migration has been started, Cloud is in read-only mode now.
The maintenance has been rescheduled from Tuesday to Monday.
for Monday 22. November 2020, 12 PM UTC / Noon (UTC time) there is a planned maintenance for the cloud.
The plan is to migrate the cloud version to the new 5.0 version. To make this happen Squidex has to run a migration of the data. The old data stays intact during this time and it should not take more than 15 minutes. During this time partial or total downtimes might happen.
How will it work?
- STEP 1: A new version will deployed, which will reject writes. Then you cannot make writes to Squidex. This prevent consistency issues.
- STEP 2: The migration will be started. This will make a copy of some of the data to new collections. Therefore the old collections will stay intact and get be used by the version deployed in STEP 1.
- STEP 3: Version 5.0 will be deployed, which will then allow writes again.
If everything goes as planned there will be no downtime except for the writes.