Repository navigation
Conversation
…erList before failover The failover walkthrough hardcoded port 30931 as the Leader, but the leader is elected at runtime and is not fixed; instruct users to find the master (BID 0) with clusterList first, and list the configured listenPorts (30911/30921/30931) to map it to a process.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #11351 +/- ##
=============================================
- Coverage 49.73% 49.71% -0.03%
+ Complexity 14348 14344 -4
=============================================
Files 1390 1390
Lines 103246 103246
Branches 13502 13502
=============================================
- Hits 51354 51328 -26
- Misses 45696 45718 +22
- Partials 6196 6200 +4 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
RockteMQ-AI
left a comment
There was a problem hiding this comment.
Review by github-manager-bot
Summary
Fixes the DLedger quick-start failover section (both EN and CN) to correctly instruct users to identify the current Leader via clusterList before killing it, instead of hardcoding port 30931 as the Leader.
Findings
- [Info]
docs/en/dledger/quick_start.md:60— The fix is correct: DLedger leader is elected at runtime and is not bound to a specific port. The original instruction could mislead users into killing a follower. - [Info]
docs/cn/dledger/quick_start.md:61— CN version is consistent with the EN change. Good. - [Info] Listing all three
listenPortvalues (30911/30921/30931) so users can mapclusterListoutput to a process is a nice touch.
Suggestions
- Minor: consider adding a short note that BID 0 means Master in the default single-cluster setup, since multi-group deployments may have multiple BID 0 entries. Not blocking.
LGTM — documentation-only change, both languages consistent, fix is accurate.
Automated review by github-manager-bot
f7fe337 to
0e6d396
Compare
|
Hi @RongtongJin, sorry for the ping — this PR has been open for a while without a review yet. The change is small and focused (identify the DLedger leader via clusterList before failover), and all tests and CI pass on my side. Would you mind taking a look when you get a chance? If it looks good, merging it would be much appreciated — happy to adjust anything if needed. Thanks a lot! |
Which Issue(s) This PR Fixes
Fixes #11334
Brief Description
The DLedger quick-start failover section (en + cn) instructed users to "kill the process bound to port 30931" as if that node were the Leader. The Leader is elected by DLedger at runtime and is not fixed — following the instruction can kill a follower, and the promised leadership switch then never happens.
The instruction now builds on the page's own earlier step: first confirm the current Leader with
mqadmin clusterList(BID 0 = Master, as the page itself explains), then kill that node's process, with the three configuredlistenPorts (30911 / 30921 / 30931 for broker-n0/n1/n2 inconf/dledger/) listed so the clusterList row can be mapped to a process, then re-check withclusterList.How Did You Test This Change?
Documentation-only change, verified statically at the PR base commit:
distribution/conf/dledger/broker-n{0,1,2}.conflistenPorts are 30911/30921/30931; "BID 为 0 的表示 Master" is the same page's existing explanation two paragraphs above; no other quick-start text depends on the hardcoded 30931-Leader assumption (grep 309in both pages).