Add cluster-config-save-behavior option to control nodes.conf save behavior#3372
Merged
enjoy-binbin merged 6 commits intoApr 10, 2026
Merged
Conversation
This commit introduces a new configuration option `cluster-persist-config` that controls how the cluster handles nodes.conf file save failures. The option supports two modes: - `sync` (default): Synchronously save the config file. If the save fails, the process exits. This maintains backward compatibility with the old behavior (before 9.1). - `best-effort`: Synchronously save the config file. If the save fails, only log a warning and continue running. This allows the node to survive disk failures (e.g., disk full, read-only filesystem) without exitting, giving administrators time to address the issue. Note that this modifies the behavior of valkey-io#1032, whereas valkey-io#1032 was "best-effort", we have now introduced a configuration option that defaults to "sync." See valkey-io#1032 discussion for more details. Background: When a disk becomes read-only or full, any cluster metadata change would trigger a nodes.conf save attempt. With the old behavior, the node would immediately exit via clusterSaveConfigOrDie(), potentially causing multiple nodes on the same machine to crash simultaneously, leading to cluster unavailability. The new `best-effort` mode addresses this by allowing nodes to continue operating even when disk writes fail. This is particularly useful in cloud environments where disk failures are more common due to scale. Note: Startup-time config saves (in clusterInit and verifyClusterConfigWithData) still use clusterSaveConfigOrDie() since disk issues at startup should cause immediate failure. Signed-off-by: Binbin <binloveplay1314@qq.com>
Signed-off-by: Binbin <binloveplay1314@qq.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## unstable #3372 +/- ##
============================================
- Coverage 74.55% 74.50% -0.05%
============================================
Files 130 130
Lines 72731 72734 +3
============================================
- Hits 54225 54192 -33
- Misses 18506 18542 +36
🚀 New features to boost your workflow:
|
Contributor
|
When we merge #2555 we can change 'best-effort' to be in bio? |
murphyjacob4
approved these changes
Mar 17, 2026
murphyjacob4
left a comment
Contributor
There was a problem hiding this comment.
Makes sense to me
We had quorum in Monday's meeting and got the verbal consensus. FYI @valkey-io/core-team in case anyone wasn't there and wants to chime in
hpatro
reviewed
Mar 17, 2026
hpatro
left a comment
Contributor
There was a problem hiding this comment.
Shall we name it as cluster-config-save-behavior ?
Member
Author
Yes, we can, it is best-effort.
Sound like a better (a more specific one) name. I am open to it. |
Contributor
|
@enjoy-binbin Shall we target it for 9.1 and get it ready ? I would personally prefer the above suggested config name (more readable / understandable by an admin). |
Signed-off-by: Binbin <binloveplay1314@qq.com>
…save Signed-off-by: Binbin <binloveplay1314@qq.com>
Signed-off-by: Binbin <binloveplay1314@qq.com>
Signed-off-by: Binbin <binloveplay1314@qq.com>
Member
Author
|
@hpatro Thanks, i changed the name to cluster-config-save-behavior, please review it again. |
hpatro
approved these changes
Apr 7, 2026
Contributor
|
Thinking about it, is nodes.conf a config file? You can't configure the cluster by changing the file. It's a snapshot of the cluster state, but not really a config file? Anyway, it's known as nodes.conf so we can still use the name |
sarthakaggarwal97
pushed a commit
to sarthakaggarwal97/valkey
that referenced
this pull request
Apr 16, 2026
…havior (valkey-io#3372) This commit introduces a new configuration option `cluster-config-save-behavior` that controls how the cluster handles nodes.conf file save failures. The option supports two modes: - `sync` (default): Synchronously save the config file. If the save fails, the process exits. This maintains backward compatibility with the old behavior (before 9.1). - `best-effort`: Synchronously save the config file. If the save fails, only log a warning and continue running. This allows the node to survive disk failures (e.g., disk full, read-only filesystem) without exitting, giving administrators time to address the issue. Note that this modifies the behavior of valkey-io#1032, whereas valkey-io#1032 was "best-effort", we have now introduced a configuration option that defaults to "sync." See valkey-io#1032 discussion for more details. Background: When a disk becomes read-only or full, any cluster metadata change would trigger a nodes.conf save attempt. With the old behavior, the node would immediately exit via clusterSaveConfigOrDie(), potentially causing multiple nodes on the same machine to crash simultaneously, leading to cluster unavailability. The new `best-effort` mode addresses this by allowing nodes to continue operating even when disk writes fail. This is particularly useful in cloud environments where disk failures are more common due to scale. Note: Startup-time config saves (in clusterInit and verifyClusterConfigWithData) still use clusterSaveConfigOrDie() since disk issues at startup should cause immediate failure. Signed-off-by: Binbin <binloveplay1314@qq.com>
madolson
pushed a commit
that referenced
this pull request
Apr 27, 2026
…havior (#3372) This commit introduces a new configuration option `cluster-config-save-behavior` that controls how the cluster handles nodes.conf file save failures. The option supports two modes: - `sync` (default): Synchronously save the config file. If the save fails, the process exits. This maintains backward compatibility with the old behavior (before 9.1). - `best-effort`: Synchronously save the config file. If the save fails, only log a warning and continue running. This allows the node to survive disk failures (e.g., disk full, read-only filesystem) without exitting, giving administrators time to address the issue. Note that this modifies the behavior of #1032, whereas #1032 was "best-effort", we have now introduced a configuration option that defaults to "sync." See #1032 discussion for more details. Background: When a disk becomes read-only or full, any cluster metadata change would trigger a nodes.conf save attempt. With the old behavior, the node would immediately exit via clusterSaveConfigOrDie(), potentially causing multiple nodes on the same machine to crash simultaneously, leading to cluster unavailability. The new `best-effort` mode addresses this by allowing nodes to continue operating even when disk writes fail. This is particularly useful in cloud environments where disk failures are more common due to scale. Note: Startup-time config saves (in clusterInit and verifyClusterConfigWithData) still use clusterSaveConfigOrDie() since disk issues at startup should cause immediate failure. Signed-off-by: Binbin <binloveplay1314@qq.com>
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
5 participants
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.
This commit introduces a new configuration option
cluster-config-save-behaviorthat controls how the cluster handles nodes.conf file save failures.
The option supports two modes:
sync(default): Synchronously save the config file. If the save fails,the process exits. This maintains backward compatibility with the old
behavior (before 9.1).
best-effort: Synchronously save the config file. If the save fails,only log a warning and continue running. This allows the node to survive
disk failures (e.g., disk full, read-only filesystem) without exitting,
giving administrators time to address the issue.
Note that this modifies the behavior of #1032, whereas #1032 was "best-effort",
we have now introduced a configuration option that defaults to "sync."
See #1032 discussion for more details.
Background:
When a disk becomes read-only or full, any cluster metadata change would
trigger a nodes.conf save attempt. With the old behavior, the node would
immediately exit via clusterSaveConfigOrDie(), potentially causing multiple
nodes on the same machine to crash simultaneously, leading to cluster
unavailability.
The new
best-effortmode addresses this by allowing nodes to continueoperating even when disk writes fail. This is particularly useful in cloud
environments where disk failures are more common due to scale.
Note: Startup-time config saves (in clusterInit and verifyClusterConfigWithData)
still use clusterSaveConfigOrDie() since disk issues at startup should cause
immediate failure.