disagg: Add exclusive label for tiflash compute node

[JaySon-Huang]

What problem does this PR solve?

Issue Number: close #9750

Problem Summary:

When creating a placement policy through tidb, tidb will create a placement-rule with "engine" notIn "tiflash" pingcap/tidb#22065
However, under tiflash disaggregated compute and storage arch, the compute node will register store with label {"engine": "tiflash_compute"}.

The logic in PD of choosing store for rule: pkg/schedule/placement/label_constraint.go @ pd

1. If the store's label.key is equal to "engine" (or "exclusive," or starts with '$'), and the rule's constraint does not contain a rule with the same label.key, then do not schedule the peer to this store.
2. Otherwise, further check if the rule's constraint matches the store's label. If it matches, schedule the peer to this store.

So the PD would pick tiflash compute node as target store to place the Region peer.

What is changed and how it works?

Add a label with {"exclusive": "no-data"}, so that PD won't select TiFlash compute node stores for placing Region

We avoid Region data being placed to TiFlash compute node in 2 ways:

• After The placement-rule generated by placement-rule-in-SQL may unexpectedly add Region peer to tiflash compute node tidb#58633, tidb will NOT set the placement rule with "engine" notIn "tiflash" explicitly.
• This PR in TiFlash compute node, it will add a label with {"exclusive": "no-data"}, so that PD won't select this store for placing Region
  • Remove --engine-role-label from proxy args, use --labels instead
  • Proxy args --labels won't overwrite the server.labels in tiflash_tikv.toml, instead, proxy append the extra labels to server.labels

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code

Side effects

• Performance regression: Consumes more CPU
• Performance regression: Consumes more Memory
• Breaking backward compatibility

Documentation

• Affects user behaviors
• Contains syntax changes
• Contains variable changes
• Contains experimental features
• Changes MySQL compatibility

Release note

Fix a bug that after executing `ALTER TABLE ... PLACEMENT POLICY ...` Region peers may be unexpectedly added to the TiFlash Compute Node in the disaggregated storage and compute architecture

[JaySon-Huang]

/hold

[JaySon-Huang]

Depends on pingcap/tidb-engine-ext#420

[JaySon-Huang]

/test pull-integration-test

[JaySon-Huang]

/unhold

The proxy change has been merged

[JaySon-Huang]

@CalvinNeo @Lloyd-Pottiger @JinheLin PTAL

[CalvinNeo]

lgtm

[ti-chi-bot]

[LGTM Timeline notifier]

Timeline:

• 2025-02-10 09:42:26.195747943 +0000 UTC m=+263188.591970004: ☑️ agreed by CalvinNeo.
• 2025-02-11 03:02:02.804762517 +0000 UTC m=+325565.200984579: ☑️ agreed by Lloyd-Pottiger.

[JaySon-Huang]

/hold

[CalvinNeo]

lgtm

[ti-chi-bot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: CalvinNeo, Lloyd-Pottiger

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:

• OWNERS [CalvinNeo,Lloyd-Pottiger]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[JaySon-Huang]

/unhold

[ti-chi-bot]

In response to a cherrypick label: failed to apply #9849 on top of branch "release-7.5":

[failed to git add conflicting files: exit status 128, failed to git commit: exit status 128]

[ti-chi-bot]

In response to a cherrypick label: failed to apply #9849 on top of branch "release-8.1":

[failed to git add conflicting files: exit status 128, failed to git commit: exit status 128]

[ti-chi-bot]

In response to a cherrypick label: failed to apply #9849 on top of branch "release-8.5":

[failed to git add conflicting files: exit status 128, failed to git commit: exit status 128]