[omdb] exhume omdb db saga show from #4378
#9478
Open
+412
−26
Conversation
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
In the days of yore, @jmpesp wrote an `omdb db saga show` command in PR #4378, which prints a trace of the execution of a given saga. Unfortunately, this code never made it to `main`. Some parts of #4378 adding _other_ saga-related OMDB commands merged in separate PRs, but the `saga show` command never made it in. Instead, when it's necessary to display the execution history of a saga for debugging, folks have maintained separate `omdb-sagas` binaries built from #4378 and copied them onto the dogfood rack or racklettes as needed for debugging. This is not an ideal situation. If an issue occurs on a customer rack rather than in development, it would be useful for support to be able to display these saga execution traces without copying a random binary onto the customer's rack, which may not be permitted depending on the customer environment. And, when the subcommand is not merged to the `main` branch and is instead maintained separately, it may drift substantially from the rest of the codebase, potentially encountering schema incompatibilities in future versions. Therefore, this PR digs up the `omdb db saga show` command from #4378 and does the minimum amount of changes necessary to make it compile with the current `main`. Almost all of the added code was written by @jmpesp, and I've just copied it and pasted it, and fixed compiler errors and such. While I'm open to addressing some of the suggestions from @dap's [review of #4378][1], I felt like the priority was to just get the code pulled forward into a working state. [1]: #4378 (review) Authored-by: James MacMahon <james@oxide.computer> Co-authored-by: Eliza Weisman <eliza@elizas.website>
hawkw
commented
Dec 5, 2025
Comment on lines
+684
to
+716
| // Copy some types from Steno, because steno uses pub(crate) everywhere. We | ||
| // don't want to change Steno to make the internals public, but these types | ||
| // should be fairly stable. | ||
| #[derive(Serialize, Deserialize)] | ||
| struct StenoDag { | ||
| pub saga_name: String, | ||
| pub graph: Graph<StenoNode, ()>, | ||
| pub start_node: NodeIndex, | ||
| pub end_node: NodeIndex, | ||
| } | ||
|
|
||
| impl StenoDag { | ||
| pub fn get(&self, node_index: NodeIndex) -> Option<&StenoNode> { | ||
| self.graph.node_weight(node_index) | ||
| } | ||
|
|
||
| pub fn get_from_saga_node_id( | ||
| &self, | ||
| saga_node_id: &nexus_db_model::saga_types::SagaNodeId, | ||
| ) -> Option<&StenoNode> { | ||
| self.get(u32::from(saga_node_id.0).into()) | ||
| } | ||
| } | ||
|
|
||
| #[derive(Serialize, Deserialize, Debug)] | ||
| enum StenoNode { | ||
| Start { params: Arc<serde_json::Value> }, | ||
| End, | ||
| Action { name: String, label: String, action_name: String }, | ||
| Constant { name: String, value: Arc<serde_json::Value> }, | ||
| SubsagaStart { saga_name: String, params_node_name: String }, | ||
| SubsagaEnd { name: String }, | ||
| } |
Member
Author
There was a problem hiding this comment.
@davepacheco left a comment on #4378 suggesting that we would be better off exposing the requisite APIs in steno, rather than duplicating these types here: #4378 (comment)
I agree with this suggestion and would quite like to address it upstream. However, I didn't think this PR ought to be blocked on waiting for those changes in Steno. Instead, I'd like to go add the requisite interfaces there and come back and change this to use them. If others disagree, I'm open to being convinced otherwise.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
In the days of yore, @jmpesp wrote an
omdb db saga showcommand in PR #4378, which prints a trace of the execution of a given saga. Unfortunately, this code never made it tomain. Some parts of #4378 adding other saga-related OMDB commands merged in separate PRs, but thesaga showcommand never made it in. Instead, when it's necessary to display the execution history of a saga for debugging, folks have maintained separateomdb-sagasbinaries built from #4378 and copied them onto the dogfood rack or racklettes as needed for debugging.This is not an ideal situation. If an issue occurs on a customer rack rather than in development, it would be useful for support to be able to display these saga execution traces without copying a random binary onto the customer's rack, which may not be permitted depending on the customer environment. And, when the subcommand is not merged to the
mainbranch and is instead maintained separately, it may drift substantially from the rest of the codebase, potentially encountering schema incompatibilities in future versions.Therefore, this PR digs up the
omdb db saga showcommand from #4378 and does the minimum amount of changes necessary to make it compile with the currentmain. Almost all of the added code was written by @jmpesp, and I've just copied it and pasted it, and fixed compiler errors and such. While I'm open to addressing some of the suggestions from @dap's review of #4378, I felt like the priority was to just get the code pulled forward into a working state.Authored-by: James MacMahon james@oxide.computer
Co-authored-by: Eliza Weisman eliza@elizas.website