feat(test runner): improve sharding algorithm to better spread similar tests among shards #30962

muhqu · 2024-05-22T15:02:25Z

Adds alternative algorithms to assign test groups to shards to better distribute tests.

Problem

Currently the way sharding works is something like this…

         [  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12]
Shard 1:  ^---------^                                      : [  1, 2, 3 ]
Shard 2:              ^---------^                          : [  4, 5, 6 ]
Shard 3:                          ^---------^              : [  7, 8, 9 ]
Shard 4:                                      ^---------^  : [ 10,11,12 ]

Tests are ordered in the way they are discovered, which is mostly alphabetically. This has the effect that test cases are sorted nearby similar tests… for example your have first 6 tests which are testing logged-in state and then 6 tests which test logged-out state. The first 6 tests require more setup time as they are testing logged-in behaviour… With the current sharding algorithm shard 1 & 2 get those slow logged-in tests and shard 3 & 4 get the more quicker tests…

Solution

This PR adds a new shardingMode configuration which allows to specify the sharding algorithm to be used…

`shardingMode: 'partition'`

That's the current behaviour, which is the default. Let me know if you have a better name to describe the current algorithm...

`shardingMode: 'round-robin'`

Distribute the test groups more evenly. It…

sorts test groups by number of tests in descending order
then loops through the test groups and assigns them to the shard with the lowest number of tests.

Here is a simple example where every test group represents a single test (e.g. --fully-parallel) ...

         [  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11, 12]
Shard 1:    ^               ^               ^              : [  1, 5, 9 ]
Shard 2:        ^               ^               ^          : [  2, 6,10 ]
Shard 3:            ^               ^               ^      : [  3, 7,11 ]
Shard 4:                ^               ^               ^  : [  4, 8,12 ]

…or a more complex scenario where test groups have different number of tests…

Original Order: [ [1], [2, 3], [4, 5, 6], [7], [8], [9, 10], [11], [12] ]
Sorted Order:   [ [4, 5, 6], [2, 3], [9, 10], [1], [7], [8], [11], [12] ]
Shard 1:           ^-----^                                                : [ [ 4,   5,   6] ]
Shard 2:                      ^--^                       ^                : [ [ 2,  3],  [8] ]
Shard 3:                              ^---^                    ^          : [ [ 9, 10], [11] ]
Shard 4:                                       ^    ^                ^    : [ [1], [7], [12] ]

`shardingMode: 'duration-round-robin'`

It's very similar to round-robin, but it uses the duration of a tests previous run as cost factor. The duration will be read from .last-run.json when available. When a test can not be found in .last-run.json it will use the average duration of available tests. When no last run info is available, the behaviour would be identical to round-robin.

Other changes

Add testDurations?: { [testId: string]: number } to .last-run.json
Add builtin lastrun reporter, which allows merge-reports to generate a .last-run.json to be generated

Appendix

Below are some runtime stats from a project I've been working on, which shows the potential benefit of this change.

The tests runs had to complete 161 tests. Single test duration ranges from a few seconds to over 2 minutes.

The partition run gives the baseline performance and illustrates the problem quite good. We have a single shard that takes almost 16 min while another one completes in under 5 min.

The round-robin algorithm gives a bit better performance, but it still has a shard that requires twice the time of another shard.

The duration-round-robin run was using the duration info from a previous run and achieves the best result by far. All shards complete in 10-11 minutes. 🏆 🎉

muhqu · 2024-05-22T15:11:53Z

Maybe it's better to make this an option to allow restoring the old behaviour. ¯_(ツ)_/¯

~~And… there should be unit-tests, no?~~ found them…

pavelfeldman · 2024-05-22T23:46:17Z

Do you think you can achieve the same better behavior with your sharding seed? Or are you looking for additional bias against subsequent tests being put into the same group?

muhqu · 2024-05-23T05:45:19Z

Do you think you can achieve the same better behavior with your sharding seed?

Not sure yet. But I will test this new sharding logic in our test setup to gather some results.

Or are you looking for additional bias against subsequent tests being put into the same group?

The seeded shuffle is basically just a quick and easy way to influence the test group to shard assignment… it's random and so it's results may vary.

However, this change is aimed to improve the sharding logic to generally yield better results, which yet needs to be proved. 😅

Currently this sharding algorithm uses the number of tests per test group as a cost metric. It would be great if we could use the test duration of a previous run (when available) to even better distribute the tests among the shards. But the algorithm would be quite similar.

pavelfeldman · 2024-05-24T15:36:32Z

I think your seed change allows users to experiment with the seeds and arrive at some better state than they are today. Any other changes w/o the timing feedback are going to yield similar results, not need to experiment with biases.

It would be great if we could use the test duration of a previous run (when available) to even better distribute the tests among the shards.

This requires a feedback loop with the test time stats which we don't have today. We recently started storing the last run stats in .last-run.json, I think it is Ok to store the test times there and to use it on the subsequent runs for better sharding, if it is available. Would you be interested in working on it?

muhqu · 2024-05-24T15:46:49Z

Yes, I would like to work on that.

I was not yet aware of the .last-run.json. Is that something that is also written by the merge reports command? Because we need the stats combined from all shard runs.

I was thinkings about adding a separate reporter for that purpose, but if those last run stats are already there…, then there might not be the need to create a separate reporter.

pavelfeldman · 2024-05-24T17:02:11Z

I was thinkings about adding a separate reporter for that purpose, but if those last run stats are already there…, then there might not be the need to create a separate reporter.

Shaping this code as reporter sounds good, but Playwright core would need to consume the output of that reporter, so it needs to be baked in. Merging those should not be a hard problem, reporter or not. Unfortunately merging mangles test ids today, so we'd need to figure that out. Maybe not using the ids altogether and falling back to the file names and test titles. Also has some tricky edge cases as tests that are fast on Chromium and are slow on Firefox...

muhqu · 2024-05-27T09:31:15Z

I've added a lastrun reporter that can be used with merge-reports to generate .last-run.json.

Surprisingly when merging the reports the test ids just had a 1 character suffix that I was able to strip off… but it doesn't feel like the right way to do this.

What's the reason to modify test ids when merging blobs? Couldn't this be done in a way that only modifies a test id when there is a collision?

muhqu · 2024-05-28T11:10:17Z

packages/playwright/src/runner/runner.ts

+  const testDurations = testRun.rootSuite?.allTests().reduce((map, t) => {
+    if (t.results.length)
+      map[t.id] = t.results.reduce((a, b) => a + b.duration, 0);
+    return map;
+  }, {} as { [testId: string]: number });


I'm actually not sure it is the right way to sum all the durations… maybe it makes more sense to calc the average? Or only include durations from successful test runs… 🤔

I think this is ok for now, we'll iterate based on feedback.

muhqu · 2024-05-28T16:32:23Z

@pavelfeldman last-run.json is a little bit tricky to work with atm… it constantly gets overwritten even if you just list tests and VS Code seems to do that from time to time to refresh the tests panel... I think it should only get written when it's actually running tests.

This reverts commit 389e571.

muhqu · 2024-05-30T15:44:05Z

@pavelfeldman […] we would probably explicitly point to the file […]

Do you have a recommendation how you would name cli parameter / configuration option?

Something like --sharding-read-last-run-info path/to/merged-last-run.json is probably too much of a mouth full? However, I think we should somehow make it obvious when data is read or written to a file…

When merging blob reports test ids are patched to make sure there is no collision when merging reports that might have overlapping test ids. However, even if you were merging reports that had no overlapping ids, all test ids will be modified, which is an undesirable side effect. This PR only modify test ids when the same test id has already been used in a previous blob report. ---- This change is also part of #30962

…icrosoft#30817)" This reverts commit 825e0e4. API review notes: sounds like this change did not solve the problem for the contributor, there is a new approach under development in microsoft#30962

lukeapage · 2024-06-22T10:39:26Z

This is awesome - we were planning on implementing something similar outside playwright before we migrate cypress tests that currently shard across 50 build servers.

Is there anything todo here I could help with to get this merged quicker ?

muhqu · 2024-06-22T12:26:44Z

Hey, I did not noticed this PR got conflicts due to the revert of my other change… I’ll resolve those conflicts in the next days. There was still on open question on whether we should introduce a command line flag to specify a last run state file or not… but I think that are improvements that can also happen after this has been merged

Though, this PR could have some more documentation changes… anyone want to help with that?

dgozman

There was still on open question on whether we should introduce a command line flag to specify a last run state file or not

We discussed this and decided that it should be fine to add --last-run-file= command line flag that will affect both the reading and writing location of the last-run.json.

Though, this PR could have some more documentation changes… anyone want to help with that?

Of course! We can help with adding more documentation after landing this PR.

dgozman · 2024-06-24T22:35:08Z

packages/playwright/src/runner/runner.ts

@@ -152,6 +152,7 @@ export class Runner {
 export type LastRunInfo = {
  status: FullResult['status'];
  failedTests: string[];
+  testDurations?: { [testId: string]: number };


We are a bit worried that this will force testId to never change to preserve backwards compatibility. Therefore, we would like to use testCase.titlePath() instead of id for now. Something like this:

export type LastRunInfo = { status: FullResult['status'], projects: { name: string, files: { path: string, tests: { title: string[], duration: number, status: TestStatus, }[], }[], }[], };

Here file.path is toPosixPath(path.relative(config.rootDir, filePath)), similar to this line.

Yes, more verbose is also better for future use-cases that will consume the last-run info. Having the titles and paths in there is more user friendly… 👍

However, removing the already existing failedTests: string[] from LastRunInfo would be a backward incompatible change… maybe we should keep it for now?

However, removing the already existing failedTests: string[] from LastRunInfo would be a backward incompatible change… maybe we should keep it for now?

There are no backwards compatibility concerns as of today, since we do not expect last-run.json to be used between different versions. It will be the case once we support "commit last-run.json to the repo for shard optimization" scenario.

That said, you can leave it as is, and we'll iterate on it.

I just tried to write the LastRunInfo in the proposed format, but its not as straight forward as I thought it would be.

That said, you can leave it as is, and we'll iterate on it.

You mean, I don't need to proceed here to get this merged? 🙏

[…] we do not expect last-run.json to be used between different versions

In our CI setup, I'm using pipeline steps to restore the last-run.json from the last successful main build. So, in our case last-run info from previous playwright versions will be used, and it would be great if there is some level of backward compatibility... at least it should not fail when the format is incorrect.

dgozman · 2024-06-24T22:36:50Z

packages/playwright/src/runner/runner.ts

+  const testDurations = testRun.rootSuite?.allTests().reduce((map, t) => {
+    if (t.results.length)
+      map[t.id] = t.results.reduce((a, b) => a + b.duration, 0);
+    return map;
+  }, {} as { [testId: string]: number });


I think this is ok for now, we'll iterate based on feedback.

github-actions · 2024-06-25T17:51:20Z

Test results for "tests 1"

7 flaky

⚠️ [chromium-library] › library/browsercontext-fetch-happy-eyeballs.spec.ts:49:3 › get should work on request fixture
⚠️ [chromium-library] › library/browsercontext-fetch-happy-eyeballs.spec.ts:49:3 › get should work on request fixture
⚠️ [chromium-library] › library/browsercontext-fetch-happy-eyeballs.spec.ts:49:3 › get should work on request fixture
⚠️ [firefox-library] › library/browsercontext-fetch-happy-eyeballs.spec.ts:49:3 › get should work on request fixture
⚠️ [chromium-library] › library/browsercontext-fetch-happy-eyeballs.spec.ts:49:3 › get should work on request fixture
⚠️ [playwright-test] › ui-mode-test-screencast.spec.ts:21:5 › should show screenshots
⚠️ [webkit-library] › library/browsercontext-fetch-happy-eyeballs.spec.ts:49:3 › get should work on request fixture

28487 passed, 649 skipped
✔️✔️✔️

Merge workflow run.

muhqu · 2024-06-25T17:55:46Z

@dgozman I've just added --last-run-file= cli flag and CI seems to be happy… any further work needed on this PR to get merged?

muhqu mentioned this pull request May 22, 2024

feat(test runner): shuffle order of tests with sharding seed #30817

Merged

This comment has been minimized.

Sign in to view

muhqu added 8 commits May 27, 2024 07:05

sharding algorithm to better spread similar tests among shards

54bbba8

improve shard algorithm by sorting test groups by number of tests

ee64b15

adjust shard.spec.ts to new algorithm

6543baf

fix reporter-json.spec.ts due to sharding

59c625e

fix reporter-junit.spec.ts due to sharding

d3b4fad

empty commit to trigger ci

4bcfc78

add test durations to last run info

061f559

allow .last-run.json to be generated via merge-reports

14ca30a

muhqu force-pushed the sharding-algorithm branch from 82809eb to 14ca30a Compare May 27, 2024 08:50

This comment has been minimized.

Sign in to view

muhqu added 2 commits May 27, 2024 15:47

Add shardingMode configuration

ce09f88

revert json/junit test changes

d42c499

This comment has been minimized.

Sign in to view

empty commit to trigger ci

b5b8174

This comment has been minimized.

Sign in to view

muhqu commented May 28, 2024

View reviewed changes

muhqu mentioned this pull request May 29, 2024

fix(runner): don't write last run info when listing tests #31062

Merged

This comment has been minimized.

Sign in to view

muhqu added 4 commits May 30, 2024 08:14

optimize: use single global set

32169b7

Feedback use outputLines

fd90424

Revert "fix(merge-reports) only change test ids when needed"

c7b58d7

This reverts commit 389e571.

Merge branch 'fix-merge-reports-test-ids' into sharding-algorithm

6cf1217

This comment has been minimized.

Sign in to view

muhqu added 3 commits May 30, 2024 10:15

empty commit to trigger ci

7dd7115

lint fix

cabfa74

Merge branch 'fix-merge-reports-test-ids' into sharding-algorithm

899c068

This comment has been minimized.

Sign in to view

Merge branch 'main' into sharding-algorithm

8aee7df

This comment has been minimized.

Sign in to view

pavelfeldman mentioned this pull request Jun 11, 2024

Revert "feat(test runner): shuffle order of tests with sharding seed … #31260

Merged

This comment has been minimized.

Sign in to view

muhqu added 2 commits June 24, 2024 15:08

Merge branch 'main' into sharding-algorithm

843d629

fix test.d.ts

2536da7

muhqu force-pushed the sharding-algorithm branch from 904a8dd to 2536da7 Compare June 24, 2024 13:11

This comment has been minimized.

Sign in to view

dgozman reviewed Jun 24, 2024

View reviewed changes

add --last-run-file CLI parameter

087a57f

muhqu requested a review from dgozman June 28, 2024 09:04

feat(test runner): improve sharding algorithm to better spread similar tests among shards #30962

Are you sure you want to change the base?

feat(test runner): improve sharding algorithm to better spread similar tests among shards #30962

Conversation

muhqu commented May 22, 2024 • edited Loading

Problem

Solution

shardingMode: 'partition'

shardingMode: 'round-robin'

shardingMode: 'duration-round-robin'

Other changes

Appendix

muhqu commented May 22, 2024 • edited Loading

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

pavelfeldman commented May 22, 2024

muhqu commented May 23, 2024 • edited Loading

pavelfeldman commented May 24, 2024

muhqu commented May 24, 2024

pavelfeldman commented May 24, 2024

This comment has been minimized.

muhqu commented May 27, 2024

This comment has been minimized.

This comment has been minimized.

Choose a reason for hiding this comment

Choose a reason for hiding this comment

muhqu commented May 28, 2024

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

muhqu commented May 30, 2024

This comment has been minimized.

lukeapage commented Jun 22, 2024

muhqu commented Jun 22, 2024

This comment has been minimized.

This comment has been minimized.

dgozman left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

github-actions bot commented Jun 25, 2024

Test results for "tests 1"

muhqu commented Jun 25, 2024

muhqu commented May 22, 2024 •

edited

Loading

`shardingMode: 'partition'`

`shardingMode: 'round-robin'`

`shardingMode: 'duration-round-robin'`

muhqu commented May 22, 2024 •

edited

Loading

muhqu commented May 23, 2024 •

edited

Loading