Bump default ScyllaDB version used in E2E's to 6.0.1 #1991
Conversation
zimnx
added
kind/feature
|
@zimnx: GitHub didn't allow me to request PR reviews from the following users: zimnx. Note that only scylladb members and repo collaborators can review this PR, and authors cannot review their own PRs. In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
scylla-operator-bot
bot
added
approved
zimnx
force-pushed
the
bump-6.0.0
branch
2 times, most recently
from
0023aa6
to
f289f83
Compare
scylla-operator-bot
bot
added
the
do-not-merge/hold
zimnx
changed the title
scylla-operator-bot
bot
added
the
do-not-merge/work-in-progress
|
New issue with our tests:
Our tests doesn't lower RF upon scaling down. |
I am fine bumping the default replicas to maintain the RF where needed.
I am not sure what you mean by that? Is there something else needed then just raising the default replicas? |
It's not matter of default replicas. In scaling test where we scaled cluster from 3->2->1->3 we didn't lower RF of keyspaces. It was a violation which started to be validated within Scylla. |
guess my point was to go say 5->4->3->4->5 with RF=3 |
This would lower case coverage compared to scenario with rf change and scaling down. But it's not what we have now, we can always add it. I'll amend node count then. |
scylla-operator-bot
bot
added
size/L
Got it, I find the case you describe valuable as well. Please make an issue and send a PR to cover RF change. |
It went well. I asked Manager team about next release, it should be out very soon. I'll unwip this PR once it's available. |
zimnx
changed the title
scylla-operator-bot
bot
removed
the
do-not-merge/work-in-progress
|
Manager was released and bumped on master so this is no longer WIP. |
fwiw manager client might need a bump as well |
|
Flake - #1525 (comment) |
Scaling E2E was aligned to make sure it doesn't break minimal required quorum on scaling changes. Existing test scaled below keyspace RF which is no longer possible, as Scylla rejects decommision when there's a keyspace having RF higher than node count. Test step checking decommission of drained node was moved earlier to fix the same quorum breakage. Alternator E2E required a table name change from which we are getting password. Table name was renamed in 6.0. Restore E2E was parametrized to make sure we test the procedure for both default ScyllaDB version and 2024.1 where a workaround explained in the documentation is required.
API wasn't changed, so it's not required. |
|
Possible flake - #1996 /retest |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: rzetelskik, tnozicka, zimnx The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
/hold cancel |
scylla-operator-bot
bot
removed
the
do-not-merge/hold
tnozicka
added
kind/machinery
Bumps default ScyllaDB version used in E2E's to 6.0.1, and align tests to 6.0 changes.
Scaling E2E was aligned to make sure it doesn't break minimal required quorum on scaling changes.
Existing test scaled below keyspace RF which is no longer possible, as Scylla rejects decommision when there's a keyspace having RF higher than node count.
Test step checking decommission of drained node was moved earlier to fix the same quorum breakage.
Alternator E2E required a table name change from which we are gettingpassword. Table name was renamed in 6.0.
Restore E2E was parametrized to make sure we test the procedure for both default ScyllaDB version and 2024.1 where a workaround explained in the documentation is required.
Fixes #1954