roachtest: fix decommission/randomize#55809
Conversation
|
This change is |
|
Draft because I stressed this locally and it's still whack. Probably will be able to spot the problem via a self-review. |
|
The problem is (I assume) that the test wants to see the decommissioned nodes in the |
The test could end up using fully decommissioned nodes for cli commands, which does not work as of cockroachdb#55286. Additionally, decommissioned nodes now become non-live after a short while, so various cli output checks had to be adjusted. Fixes cockroachdb#55581. Release note: None
|
Finally good to go, got 50 consecutive iterations in. |
| // Partially decommission then recommission n1, from another | ||
| // random node. Run a couple of status checks while doing so. | ||
| // We hard-code n1 to guard against the hypothetical case in which | ||
| // targetNode has no replicas (yet) to begin with, which would make |
There was a problem hiding this comment.
I wonder how long till we'll want to to support a way for operators to only mark a node as decommissioning (and not fully decommissioned). Then we'll have come full circle.
|
TFTR! bors r=irfansharif |
|
Build failed (retrying...): |
|
Build succeeded: |
The test could end up using fully decommissioned nodes for cli commands,
which does not work as of #55286.
Fixes #55581.
Release note: None