Alternative backup and restore approach

[ingria]

Problem Statement

I've been using sentry for years, and during that time I've had multiple times when I had to move my sentry instance to another server. And every time something doesn't go right and I end up installing and setting up new instance from scratch.

Here're the examples just from the last couple of months: #2424 #2524

I'm not alone with such a problems: a simple search for "backup" or "restore" on github shows a lot of issues (#2366, #2259, #2468 - that's just on the first page of the search results).

Solution Brainstorm

Maybe some other backup/restore approach would be better? Like copying all docker images and volumes across different machines?

[azaslavsky]

We've been working hard on improving this workflow (we know it was broken prior to 23.9.1), and have released a number of improvements in this area (see the release notes for 23.9.1 and 23.10.0 for more). Can you try using the script from the latest version of Sentry (23.10.1), and reporting back if it still breaks for you? That's an important data point for us, and we're eager to make this better if there are still issues!

[ingria]

Just created a fresh installation specifically to test backup/restore. Both versions are master (web ui shows Sentry 23.11.0.dev 0ea10232).

Now I have this error:

RpcImportError(
    kind: Unknown,
    on: InstanceID(model: 'sentry.userip'),
    reason: Unknown internal error occurred: get() returned more than one UserIP -- it returned 2!
)

We've been working hard on improving this workflow

Sure, I'm just suggesting to add some alternative solution besides the backup/restore scripts.

[azaslavsky]

This is a duplicate of #2468. This is a known issue I am actively working on - I expect to have a fix up later this week, or perhaps early next if I stumble on something unexpected.

Sure, I'm just suggesting to add some alternative solution besides the backup/restore scripts.

There is one: backing up the full volumes, as described in the docs