search

microsoft / PlanetaryComputer

Issues, discussions, and information about the Microsoft Planetary Computer
https://planetarycomputer.microsoft.com/
MIT License
185 stars 8 forks source link

Hub downtime today (May 24th) #352

Closed TomAugspurger closed 6 months ago

TomAugspurger commented 6 months ago

As mentioned in https://github.com/microsoft/PlanetaryComputer/discussions/347, we'll be moving the DNS that the Hub is served under to https://planetarycomputer-hub.microsoft.com. The hub's officially documented URL https://planetarycomputer.microsoft.com/compute will redirect to its new home. If you bookmarked the pccompute.westeurope.cloudapp.azure.com URL then you'll want to update to the new one.

As part of this migration, we'll have a few periods of downtime where you won't be able to log in. I'll start the migration at about 5:00 PM US/Central today, May 24th.

https://planetarycomputer-hub.microsoft.com will be the domain for the Hub till its retirement.

TomAugspurger commented 6 months ago

Deploying https://github.com/microsoft/planetary-computer-hub/pull/108 shortly. This will prevent users from logging in but shouldn't affect any running notebook servers.

TomAugspurger commented 6 months ago

That deployment is done.

Next up is https://github.com/microsoft/planetary-computer-hub/pull/104, which will again prevent users from logging on while the Hub pod is restarting.

TomAugspurger commented 6 months ago

Final deployment is for https://github.com/microsoft/planetary-computer-hub/pull/109. This will have some longer downtime (~20-30 minutes) as I bring up Application Gateway. This is what will flip the DNS entry to https://planetarycomputer-hub.microsoft.com/.

This might disrupt currently running notebook servers. I'm not sure.

Edit: as expected, https://pccompute.westeurope.cloudapp.azure.com/ is now erroring. I'll be bringing the Hub back up under the new URL as soon as possible.

Edit: bringing up application gateway.

TomAugspurger commented 6 months ago

Application Gateway is up. Now it's doing it's thing with ingresses before the hub will be accessible.

TomAugspurger commented 6 months ago

Whoops, misconfigured the ingress (copy-pasted a staging name into prod). Fixed in https://github.com/microsoft/planetary-computer-hub/pull/111 and deploying that now.

TomAugspurger commented 6 months ago

Getting some "Backend Unhealthy" errors, and https://planetarycomputer-hub.microsoft.com/ is throwing 500s unexpectedly. Investigating.

TomAugspurger commented 6 months ago

Cycling the appgw pod did it (or I just needed to wait a bit longer).

We should be up at https://planetarycomputer-hub.microsoft.com/. https://planetarycomputer.microsoft.com/compute should redirect there as well, so if you were using that URL previously you shouldn't have any changes (other than where you get redirected to).

Ping me if you have any issues.