chore: release v0.0.28 (state-commit write-mode pin) - #53
Merged
Conversation
Carries #51, which adds storage.state_commit.write_mode_enable_auto, and #52, which fixes two pointer-resolution defects the review of #51 surfaced. The key matters because the binary reads sc-write-mode-enable-auto under [state-commit] to decide whether to derive its write mode from committed migration state or honor the configured write_mode. The unified schema had no field for it, so BuildRegistry never produced the key and ApplyOverrides rejected it as unknown: a node could not be pinned to an explicit write mode through the controller at all. Editing app.toml on the volume works only until the next start, when the sidecar renders the file again from the intent. The field is *bool because all three states are distinct. Nil renders no key and leaves the binary's own default in force, which is what every existing node needs; a bare bool cannot express the pin, since omitempty drops an explicit false and its absence makes every node emit the key. #52 is the reason to take both together. Resolving a pointer field by reusing whatever pointee was already there produced two failures. ResolveIncrementalIntent shallow-copies the caller's config, so overriding a non-nil pointer wrote through the shared pointee and silently changed the caller's config — which diff, retry, and rollback logic then read as the pre-override state. And allocating before parsing left a rejected value behind as a pointer to the zero value: for this key that zero is false, which renders the pin. ResolveEnv only warns on a bad value and keeps going, so a typo in an env var could have taken a node off a governance-driven migration with nothing surfaced. Consumed next by sei-k8s-controller, whose sidecar resolves overrides during config-apply and is therefore what makes this real on a running node. The controller pins v0.0.27 today. Releasing changes no node's behaviour. write_mode_enable_auto defaults to nil, which renders no key. The pointer fix also makes consensus.unsafe_bypass_commit_timeout_override settable through an override for the first time; no existing node can carry it, because setting it failed before. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
PR SummaryLow Risk Overview There are no code or config schema changes in this diff—only the version string used to tag and consume this module (e.g. by downstream controllers that pin the sidecar). Reviewed by Cursor Bugbot for commit 7c2803a. Bugbot is set up for automated code reviews on this repo. Configure here. |
|
Suggested version: Changes in (empty)
Automatically created GitHub ReleaseA draft GitHub Release has been created. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bumps
version.jsontov0.0.28. One line; no source change.What it carries
storage.state_commit.write_mode_enable_autoWhy the key matters
The binary reads
sc-write-mode-enable-autounder[state-commit]to decide whether to derive its write mode from committed migration state or honor the configuredwrite_mode. The unified schema had no field for it, soBuildRegistrynever produced the key andApplyOverridesrejected it as unknown — a node could not be pinned to an explicit write mode through the controller at all. Editingapp.tomlon the volume works only until the next start, when the sidecar renders the file again from the intent.The field is
*boolbecause all three states are distinct. Nil renders no key and leaves the binary's default in force, which is what every existing node needs. A bareboolcannot express the pin: withomitemptyan explicitfalseis dropped, and without it every node in the fleet starts emitting the key.Why #52 ships with it
Resolving a pointer field by reusing whatever pointee was already there produced two failures:
Aliasing the caller.
ResolveIncrementalIntentshallow-copies the caller's config, so the copy and the original share every pointer. Overriding a non-nil pointer wrote through the shared pointee and silently changed the caller's config — which diff, retry, and rollback logic then read as the pre-override state.A rejected value becoming a pin. Allocating before parsing left a rejected value behind as a pointer to the zero value. For this key that zero is
false, which renderssc-write-mode-enable-auto = false— the pin.ResolveEnvonly warns on a bad value and keeps going, so a typo in an env var could have taken a node off a governance-driven migration with nothing surfaced.The value now parses into a fresh pointee and the pointer is replaced only on success.
Testing
make testgo vet ./...gofmt -l .staticcheck ./...Ran the tests covering both changes specifically:
TestWriteModeEnableAuto_Tristate— unset omits the key, override to false renders false, override to true renders trueTestWriteModeEnableAuto_RoundTripTestApplyOverrides_PointerDoesNotAliasCallerTestApplyOverrides_PointerRejectedValueLeavesFieldUnsetAll pass. The tristate behaviour is what the "changes no node's behaviour" claim below rests on.
Behaviour
Releasing changes no node's behaviour.
write_mode_enable_autodefaults to nil, which renders no key.The pointer fix also makes
consensus.unsafe_bypass_commit_timeout_overridesettable through an override for the first time. No existing node can carry it, because setting a pointer field failed before #51.Consumed next
sei-k8s-controller, whose sidecar resolves overrides during config-apply and is therefore what makes this real on a running node. It pinsv0.0.27today, across all three modules.🤖 Generated with Claude Code