Keep hybrid DLS safe across mixed versions - #6451
Conversation
Signed-off-by: Cédric Pelvet <cedric.pelvet@gmail.com>
PR Reviewer Guide 🔍(Review updated until commit 81a742a)Here are some key observations to aid the review process:
|
PR Code Suggestions ✨Latest suggestions up to 81a742a Explore these optional code suggestions:
Previous suggestionsSuggestions up to commit beec2a1
Suggestions up to commit c237a53
|
Signed-off-by: Cédric Pelvet <cedric.pelvet@gmail.com>
|
Persistent review updated to latest commit beec2a1 |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #6451 +/- ##
==========================================
+ Coverage 75.66% 75.79% +0.13%
==========================================
Files 456 457 +1
Lines 30414 30508 +94
Branches 4605 4615 +10
==========================================
+ Hits 23012 23123 +111
+ Misses 5281 5267 -14
+ Partials 2121 2118 -3
🚀 New features to boost your workflow:
|
We don't need to do this if this code hasn't been released yet |
I think we do to support mixed mode because the coordinator can either be an unpatched or a patched node. Here is the analysis of GPT5.6 Daybreak Blue:
|
|
@sharp-pixel that would matter if your last PR (#6416) was in a released version, but its only in mainline and unreleased. Given that, bwc checks like this block: Can be simplified to: since we don't actually check the value in that header, only that its not-null (or having the value of "true"). |
Signed-off-by: Cédric Pelvet <cedric.pelvet@gmail.com>
|
Persistent review updated to latest commit 81a742a |
d39e592
into
opensearch-project:main
Summary
- Continue accepting the legacy sentinel from unpatched nodes during rolling upgrades.Security review
No security finding under the mixed patched-node and unpatched-node cluster assumption.