dns: fix constant 30s backoff for re-resolution #7262

purnesh42H · 2024-05-23T14:51:58Z

Fixes #7186 with a test

RELEASE NOTES:

dns: fix bug that was causing constant 30s backoff between every resolution request

purnesh42H · 2024-05-23T15:02:53Z

Without Fix

1st resolve sent 2024-05-23 22:01:10.566367 +0530 IST m=+0.046273376 /Users/purneshdixit/Google/Dev/grpc-go/internal/resolver/dns/resolver.go:49: testutils.ResolverClientConn: UpdateState({ "Addresses": [ { "Addr": "1.2.3.4:443", "ServerName": "", "Attributes": null, "BalancerAttributes": null, "Metadata": null } ], "Endpoints": null, "ServiceConfig": { "Config": { "Config": null, "Methods": {} }, "Err": null }, "Attributes": null }) 1st resolve completed 2024-05-23 22:01:15.568635 +0530 IST m=+5.048577209

2nd resolve sent 2024-05-23 22:01:20.569472 +0530 IST m=+10.049450126 /Users/purneshdixit/Google/Dev/grpc-go/internal/resolver/dns/resolver.go:49: testutils.ResolverClientConn: UpdateState({ "Addresses": [], "Endpoints": null, "ServiceConfig": { "Config": { "Config": null, "Methods": {} }, "Err": null }, "Attributes": null }) 2nd resolve completed 2024-05-23 22:01:25.571348 +0530 IST m=+15.051360709

With Fix

1st resolve sent 2024-05-23 22:09:39.675797 +0530 IST m=+0.048168126 /Users/purneshdixit/Google/Dev/grpc-go/internal/resolver/dns/resolver.go:49: testutils.ResolverClientConn: UpdateState({ "Addresses": [ { "Addr": "1.2.3.4:443", "ServerName": "", "Attributes": null, "BalancerAttributes": null, "Metadata": null } ], "Endpoints": null, "ServiceConfig": { "Config": { "Config": null, "Methods": {} }, "Err": null }, "Attributes": null }) 1st resolve completed 2024-05-23 22:09:44.677399 +0530 IST m=+5.049806168

2nd resolve sent 2024-05-23 22:09:49.678512 +0530 IST m=+10.050954668 /Users/purneshdixit/Google/Dev/grpc-go/internal/resolver/dns/resolver.go:49: testutils.ResolverClientConn: UpdateState({ "Addresses": [], "Endpoints": null, "ServiceConfig": { "Config": { "Config": null, "Methods": {} }, "Err": null }, "Attributes": null }) 2nd resolve completed 2024-05-23 22:09:49.679501 +0530 IST m=+10.051943126

sebastianjonasson · 2024-05-28T08:40:41Z

I've tested this and can confirm that it resolves the issue described in #7186.

internal/resolver/dns/dns_resolver_test.go

purnesh42H · 2024-05-28T09:45:36Z

TestMinResolutionInterval_NoExtraDelay fails for main branch code but passes for the change. No failures in 100K forge

easwars · 2024-05-28T17:52:19Z

@purnesh42H : This PR definitely needs a release note because we are fixing a bug that we introduced.

internal/resolver/dns/dns_resolver_test.go

easwars · 2024-05-28T18:13:17Z

internal/resolver/dns/dns_resolver_test.go

+// the likelihood of flakiness due to timing issues.
+for i := 0; i < 4; i++ {
+verifyUpdateFromResolver(ctx, t, stateCh, wantAddrs, nil, wantSC)
+time.Sleep(1 * time.Second) // respect resolution rate of 1 second for re-resolve


We try to avoid sleeping in our tests as much as possible. Sleeping causes the tests to unnecessarily flaky, and run for longer than actually required.

Instead, if you override internal.TimeAfterFunc, you can control exactly when this function pushes something on the channel read by the code. This way, you can avoid sleeping as well.

There is currently an overrideTimeAfterFunc helper function, but this sets the after func to be invoked after a set duration. Instead you can override internal.TimeAfterFunc to return a channel that is controlled by the test, and have it push to it whenever it wants the code under test to be unblocked.

I don't think we should overriding internal.TimeAfterFunc because we want to test the actual implementation work correctly with MinResolutionInterval. Here, we just want to wait outside for 1 second before making another request so I just replaced time.Sleep with another channel using TimeAfterFunc

No, we don't want this test to run for 4+ seconds.

Instead I suggest the following:

Define a variable for internal.TimeNowFunc, just like how it is done for internal.TimeAfterFunc

set this variable to time.Now in the init function

Override this variable to return a time value that you control in your test

Override internal.TimeAfterFunc in your test and verify that it is called with the expected duration

Let me know if this doesn't work.

Thanks for the suggestion. Modified the test to override time.Until to calculate remaining wait to allow resolution after 1s i.e. equivalent of making re-resolution after 1s and verify the expected wait duration is very small.

internal/resolver/dns/dns_resolver_test.go

easwars · 2024-06-04T15:58:17Z

internal/resolver/dns/dns_resolver_test.go

 return durChan, timeChan
 }

+// Override the remaining time used by the DNS resolver to allow resolution


https://google.github.io/styleguide/go/decisions#comment-sentences

easwars · 2024-06-04T15:58:39Z

internal/resolver/dns/dns_resolver_test.go

+// between two resolution requests apart from [minResolutionInterval].
+// It sets the minResolutionInterval 1s and overrides timeUntilFunc to
+// calculate remaining time to allow resolution after 1s and verifies that
+// remaining time to allow resolution is very small


Nit: Same here. Please terminate with a period.

easwars · 2024-06-04T15:59:28Z

internal/resolver/dns/dns_resolver_test.go

+
+// TestMinResolutionInterval_NoExtraDelay verifies that there is no extra delay
+// between two resolution requests apart from [minResolutionInterval].
+// It sets the minResolutionInterval 1s and overrides timeUntilFunc to


Nit: Switch It sets the minResolutionInterval 1s with either It sets the minResolutionInterval to 1s or It sets a minResolutionInterval of 1s

easwars · 2024-06-04T17:15:02Z

internal/resolver/dns/dns_resolver_test.go

+ctx, cancel := context.WithTimeout(context.Background(), defaultTestTimeout)
+defer cancel()


Please don't recreate the test context. Use the one created earlier. Using the same top-level test context will ensure that all operations in the test complete within defaultTestTimeout.

Done. Missed removing it after getting the test work

easwars · 2024-06-04T17:30:25Z

internal/resolver/dns/dns_resolver_test.go

+},
+}
+overrideNetResolver(t, tr)
+var minResolutionInterval = 1 * time.Second


Use const instead.

internal/resolver/dns/dns_resolver_test.go

codecov · 2024-06-05T19:16:21Z

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.49%. Comparing base (adf976b) to head (6d8cc1b).
Report is 86 commits behind head on master.

❗ Current head 6d8cc1b differs from pull request most recent head deff579

Please upload reports for the commit deff579 to get more accurate results.

Additional details and impacted files

@@ Coverage Diff @@ ## master #7262 +/- ## ========================================== - Coverage 81.24% 80.49% -0.75%  ========================================== Files 345 348 +3 Lines 33941 33985 +44 ========================================== - Hits 27574 27357 -217  - Misses 5202 5432 +230  - Partials 1165 1196 +31

Files	Coverage Δ
internal/resolver/dns/dns_resolver.go	`89.39% <100.00%> (+0.08%)`	⬆️

... and 49 files with indirect coverage changes

purnesh42H force-pushed the dns-resolver-backoff-fix branch from efa9843 to 074c279 Compare May 23, 2024 16:47

purnesh42H added the Type: Bug label May 27, 2024

purnesh42H modified the milestone: 1.65 Release May 27, 2024

purnesh42H marked this pull request as ready for review May 27, 2024 20:20

purnesh42H force-pushed the dns-resolver-backoff-fix branch 5 times, most recently from dbbbea6 to 7484d21 Compare May 28, 2024 04:08

purnesh42H requested review from aranjans and arjan-bal May 28, 2024 04:08

aranjans added this to the 1.65 Release milestone May 28, 2024

arjan-bal reviewed May 28, 2024

View reviewed changes

internal/resolver/dns/dns_resolver_test.go Outdated Show resolved Hide resolved

internal/resolver/dns/dns_resolver_test.go Show resolved Hide resolved

internal/resolver/dns/dns_resolver_test.go Outdated Show resolved Hide resolved

purnesh42H assigned easwars May 28, 2024

easwars reviewed May 28, 2024

View reviewed changes

easwars assigned purnesh42H and unassigned easwars May 28, 2024

easwars removed the request for review from aranjans May 28, 2024 18:18

purnesh42H force-pushed the dns-resolver-backoff-fix branch 2 times, most recently from ce604c2 to e5f053b Compare May 29, 2024 09:57

purnesh42H requested review from arjan-bal and easwars May 29, 2024 09:57

purnesh42H assigned easwars and unassigned purnesh42H May 29, 2024

purnesh42H force-pushed the dns-resolver-backoff-fix branch from e5f053b to 78a490c Compare May 30, 2024 05:13

Replace time.Sleep with channel using time.After

16c904d

purnesh42H force-pushed the dns-resolver-backoff-fix branch from bec24bd to 23dd2e8 Compare June 4, 2024 08:22

purnesh42H assigned easwars and unassigned purnesh42H Jun 4, 2024

purnesh42H requested a review from easwars June 4, 2024 08:25

purnesh42H force-pushed the dns-resolver-backoff-fix branch 4 times, most recently from e6304b1 to 41e46e4 Compare June 4, 2024 11:36

Modified TestMinResolutionInterval_NoExtraDelay to avoid running for 4s

a775921

purnesh42H force-pushed the dns-resolver-backoff-fix branch from 41e46e4 to a775921 Compare June 4, 2024 11:44

easwars reviewed Jun 4, 2024

View reviewed changes

easwars assigned purnesh42H and unassigned easwars Jun 4, 2024

purnesh42H force-pushed the dns-resolver-backoff-fix branch from 23d7528 to 6d8cc1b Compare June 5, 2024 19:12

purnesh42H requested a review from easwars June 5, 2024 19:12

purnesh42H assigned easwars and unassigned purnesh42H Jun 5, 2024

Addressed comment style and test improvement as suggested

deff579

purnesh42H force-pushed the dns-resolver-backoff-fix branch from 6d8cc1b to deff579 Compare June 5, 2024 19:19

arvindbr8 modified the milestones: 1.65 Release, 1.66 Release Jun 6, 2024

easwars approved these changes Jun 6, 2024

View reviewed changes

easwars merged commit 5a289d9 into grpc:master Jun 6, 2024
1 check passed

purnesh42H added a commit to purnesh42H/grpc-go that referenced this pull request Jun 7, 2024

dns: fix constant 30s backoff for re-resolution (grpc#7262)

4ad7d60

purnesh42H mentioned this pull request Jun 7, 2024

dns: fix constant 30s backoff for re-resolution (#7262) #7311

Merged

easwars pushed a commit that referenced this pull request Jun 7, 2024

dns: fix constant 30s backoff for re-resolution (#7262) (#7311)

9b970fd

tbg mentioned this pull request Nov 27, 2024

DEPS: upgrade grpc to v1.68.0 cockroachdb/cockroach#136278

Closed

16 tasks

github-actions bot locked as resolved and limited conversation to collaborators Dec 4, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

dns: fix constant 30s backoff for re-resolution #7262

dns: fix constant 30s backoff for re-resolution #7262

Uh oh!

purnesh42H commented May 23, 2024 •

edited

Loading

purnesh42H commented May 23, 2024 •

edited

Loading

sebastianjonasson commented May 28, 2024

Uh oh!

Uh oh!

Uh oh!

purnesh42H commented May 28, 2024 •

edited

Loading

easwars commented May 28, 2024

Uh oh!

easwars May 28, 2024

purnesh42H May 29, 2024 •

edited

Loading

easwars May 31, 2024

purnesh42H Jun 4, 2024 •

edited

Loading

Uh oh!

easwars Jun 4, 2024

easwars Jun 4, 2024

easwars Jun 4, 2024

easwars Jun 4, 2024

purnesh42H Jun 5, 2024

easwars Jun 4, 2024

purnesh42H Jun 5, 2024

Uh oh!

codecov bot commented Jun 5, 2024 •

edited

Loading

Uh oh!

Labels

6 participants

		ctx, cancel := context.WithTimeout(context.Background(), defaultTestTimeout)
		defer cancel()

dns: fix constant 30s backoff for re-resolution #7262

dns: fix constant 30s backoff for re-resolution #7262

Uh oh!

Conversation

purnesh42H commented May 23, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

purnesh42H commented May 23, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

sebastianjonasson commented May 28, 2024

Uh oh!

Uh oh!

Uh oh!

purnesh42H commented May 28, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

easwars commented May 28, 2024

Uh oh!

Choose a reason for hiding this comment

purnesh42H May 29, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Choose a reason for hiding this comment

purnesh42H Jun 4, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

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

Uh oh!

codecov bot commented Jun 5, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Labels

6 participants

purnesh42H commented May 23, 2024 •

edited

Loading

purnesh42H commented May 23, 2024 •

edited

Loading

purnesh42H commented May 28, 2024 •

edited

Loading

purnesh42H May 29, 2024 •

edited

Loading

purnesh42H Jun 4, 2024 •

edited

Loading

codecov bot commented Jun 5, 2024 •

edited

Loading