Update Dr.CI to use the list of unstable issues (#5131)

This is the next part of #5122
where Dr.CI also uses the list of unstable issues to check if a job is
an unstable job. The same function `isUnstableJob` that powers HUD from
#5122 is used.

### Testing

Mark a job on ET as unstable
pytorch/executorch#3344 as shown on HUD
https://hud.pytorch.org/hud/pytorch/executorch/main and Dr.CI applies
the same logic on a PR pytorch/executorch#3318
to show the job as unstable.

```
curl --request POST \
--url "http://localhost:3000/api/drci/drci?prNumber=3318" \
--header "Authorization: TOKEN" \
--data 'repo=executorch'
```

<!-- drci-comment-start -->

## 🔗 Helpful Links
### 🧪 See artifacts and rendered test results at
[hud.pytorch.org/pr/pytorch/executorch/3318](https://hud.pytorch.org/pr/pytorch/executorch/3318)
* 📄 Preview [Python docs built from this
PR](https://docs-preview.pytorch.org/pytorch/executorch/3318/index.html)

Note: Links to docs will display an error until the docs builds have
been completed.


## ✅ You can merge normally! (1 Unrelated Failure)
As of commit f712e381c161901b733baa6b1fe7d85dc25404d3 with merge base
b669056c1cff5f7fe3786df9e68a14447cd5410b (<sub><sub><img alt="image"
width=70
src="https://img.shields.io/date/1713981882?label=&color=FFFFFF&style=flat-square"></sub></sub>):
<details ><summary><b>UNSTABLE</b> - The following job failed but was
likely due to flakiness present on trunk and has been marked as
unstable:</summary><p>

* [Android / test-llama-app / mobile-job
(android)](https://hud.pytorch.org/pr/pytorch/executorch/3318#24225451316)
([gh](https://github.com/pytorch/executorch/actions/runs/8821313480/job/24225451316))
`Credentials could not be loaded, please check your action inputs: Could
not load credentials from any providers`
</p></details>


This comment was automatically generated by Dr. CI and updates every 15
minutes.
<!-- drci-comment-end -->

Author: huydhn

torchci/pages/api/drci/drci.ts

@@ -4,7 +4,7 @@ import {
   fetchRecentWorkflows,
   fetchFailedJobsFromCommits,
 } from "lib/fetchRecentWorkflows";
-import { RecentWorkflowsData } from "lib/types";
+import { RecentWorkflowsData, IssueData } from "lib/types";
 import {
   NUM_MINUTES,
   formDrciComment,
@@ -29,6 +29,7 @@ import {
   isSameFailure,
   removeCancelledJobAfterRetry,
   backfillMissingLog,
+  isUnstableJob,
 } from "lib/jobUtils";
 import getRocksetClient from "lib/rockset";
 import _ from "lodash";
@@ -92,6 +93,7 @@ export async function updateDrciComments(
   await addMergeBaseCommits(octokit, repo, head, workflowsByPR);
   const sevs = getActiveSEVs(await fetchIssuesByLabel("ci: sev"));
   const flakyRules: FlakyRule[] = (await fetchJSON(FLAKY_RULES_JSON)) || [];
+  const unstableIssues: IssueData[] = await fetchIssuesByLabel("unstable");
   const baseCommitJobs = await getBaseCommitJobs(workflowsByPR);
   const existingDrCiComments = await getExistingDrCiComments(
     `${OWNER}/${repo}`,
@@ -117,7 +119,8 @@ export async function updateDrciComments(
           pr_info,
           flakyRules,
           baseCommitJobs.get(pr_info.merge_base) || new Map(),
-          labels || []
+          labels || [],
+          unstableIssues || []
         );
 
       failures[pr_info.pr_number] = {
@@ -628,7 +631,8 @@ export async function getWorkflowJobsStatuses(
   prInfo: PRandJobs,
   flakyRules: FlakyRule[],
   baseJobs: Map<string, RecentWorkflowsData[]>,
-  labels: string[] = []
+  labels: string[] = [],
+  unstableIssues: IssueData[] = []
 ): Promise<{
   pending: number;
   failedJobs: RecentWorkflowsData[];
@@ -649,7 +653,7 @@ export async function getWorkflowJobsStatuses(
     ) {
       pending++;
     } else if (job.conclusion === "failure" || job.conclusion === "cancelled") {
-      if (job.name !== undefined && job.name.includes("unstable")) {
+      if (await isUnstableJob(job, unstableIssues)) {
         unstableJobs.push(job);
       } else if (isBrokenTrunk(job, baseJobs)) {
         brokenTrunkJobs.push(job);

torchci/test/drci.test.ts

@@ -231,6 +231,20 @@ const unstableA = getDummyJob({
   runnerName: "dummy",
 });
 
+// From the list of mock unstable jobs
+const unstableB = getDummyJob({
+  name: "trunk / test-coreml-delegate / macos-job",
+  conclusion: "failure",
+  completed_at: "2022-07-13T19:34:03Z",
+  html_url: "a",
+  head_sha: "abcdefg",
+  id: "1",
+  pr_number: 1001,
+  failure_lines: ["a", "b"],
+  failure_captures: ["a", "b"],
+  runnerName: "dummy",
+});
+
 const sev: IssueData = {
   number: 85362,
   title: "docker pulls failing with no space left on disk",
@@ -528,23 +542,37 @@ describe("Update Dr. CI Bot Unit Tests", () => {
   });
 
   test("test flaky, broken trunk, and unstable jobs are filtered out", async () => {
-    const originalWorkflows = [failedA, failedB, unstableA];
+    const originalWorkflows = [failedA, failedB, unstableA, unstableB];
     const workflowsByPR = await updateDrciBot.reorganizeWorkflows(
       "pytorch",
       "pytorch",
       originalWorkflows
     );
+    const mockUnstableIssues: IssueData[] = [
+      {
+        number: 3264,
+        title: "UNSTABLE trunk / test-coreml-delegate / macos-job",
+        html_url: "https://github.com/pytorch/executorch/issues/3264",
+        state: "open",
+        body: "",
+        updated_at: "2024-04-24T00:44:19Z",
+        author_association: "CONTRIBUTOR",
+      },
+    ];
     const pr_1001 = workflowsByPR.get(1001)!;
+
     const { failedJobs, brokenTrunkJobs, flakyJobs, unstableJobs } =
       await updateDrciBot.getWorkflowJobsStatuses(
         pr_1001,
         [{ name: failedB.name!, captures: failedB.failure_captures }],
-        new Map().set(failedA.name, [failedA])
+        new Map().set(failedA.name, [failedA]),
+        [],
+        mockUnstableIssues
       );
     expect(failedJobs.length).toBe(0);
     expect(brokenTrunkJobs.length).toBe(1);
     expect(flakyJobs.length).toBe(1);
-    expect(unstableJobs.length).toBe(1);
+    expect(unstableJobs.length).toBe(2);
   });
 
   test(" test flaky rule regex", async () => {

torchci/test/drciBot.test.ts

@@ -174,6 +174,8 @@ describe("verify-drci-functionality", () => {
       .reply(200, { results: [] })
       .post((url) => url.includes("commit_failed_jobs"))
       .reply(200, { results: [] })
+      .post((url) => url.includes("issue_query"))
+      .reply(200, { results: [] })
       .post(
         (url) => url.includes("self/queries"),
         (body) => JSON.stringify(body).includes("merge_base_commit_date")