server: make the span stats fan-out more fault tolerant #108456

zachlite · 2023-08-09T16:11:28Z

This commit adds improved fault tolerance to the span stats fan-out:

Errors encountered during the fan-out will not invalidate the
entire request. Now, a span stats fan-out will always return a
roachpb.SpanStatsResponse that has been updated by values from nodes that
service their requests without error. In the extreme case where there's
a failure encountered on every node, an empty response is returned.

Errors that are encountered are logged, and then appended to the response
in the newly added Errors field.
Nodes must service requests within the timeout passed to iterateNodes.
For span stats, the value comes from a new cluster setting:
'server.span_stats.node.timeout', with a default value of 1 minute.

Resolves #106097
Epic: none
Release note (ops change): Span stats requests will return a partial
result if the request encounters any errors. Errors that would have
previously terminated the request are now included in the response.

cockroach-teamcity · 2023-08-09T16:11:40Z

This change is

j82w

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @nkodali and @zachlite)

-- commits line 9 at r1:
What is the reason for removing the error from the response? The error in the response allows users to see what the issue is and to avoids ambiguity with a node missing.

j82w

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @nkodali and @zachlite)

-- commits line 11 at r1:
How about? Errors that are encountered are logged, and also returned on a property in the response along with the successful requests.

pkg/roachpb/span_stats.go line 41 at r1 (raw file):

	time.Minute,
	settings.NonNegativeDuration,
).WithPublic()

Does this really need to be public? We shouldn't make cluster setting public unless we expect users to change it.

zachlite

I've updated the response to include the errors encountered on each node. I've also updated the test to make sure that in the extreme case of failure on each node, stats for each span requested is still accessible.

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @j82w and @nkodali)

-- commits line 11 at r1:

Previously, j82w (Jake) wrote…

How about? Errors that are encountered are logged, and also returned on a property in the response along with the successful requests.

Done.

pkg/roachpb/span_stats.go line 41 at r1 (raw file):

Previously, j82w (Jake) wrote…

Does this really need to be public? We shouldn't make cluster setting public unless we expect users to change it.

Done.

knz

Reviewed 8 of 14 files at r1, 5 of 7 files at r2, all commit messages.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @j82w, @nkodali, and @zachlite)

pkg/server/span_stats_server.go line 137 at r2 (raw file):

	}

	res.Errors = strings.Join(errorMessages, ",")

Not in love with this. Why not make the proto a repeated string errors?

zachlite

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @j82w, @knz, and @nkodali)

pkg/server/span_stats_server.go line 137 at r2 (raw file):

Previously, knz (Raphael 'kena' Poss) wrote…

Not in love with this. Why not make the proto a repeated string errors?

You're right. Done!

maryliag

Reviewed 6 of 14 files at r1, 2 of 7 files at r2, 4 of 5 files at r3, all commit messages.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @j82w, @knz, @nkodali, and @zachlite)

pkg/server/span_stats_test.go line 200 at r3 (raw file):

	defer log.Scope(t).Close(t)
	ctx := context.Background()
	const numNodes = 5

For the test to be more complete, at least of of the nodes should return something valid, to make sure that it's working and we get some info.

zachlite

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @j82w, @knz, @maryliag, and @nkodali)

pkg/server/span_stats_test.go line 200 at r3 (raw file):

Previously, maryliag (Marylia Gutierrez) wrote…

For the test to be more complete, at least of of the nodes should return something valid, to make sure that it's working and we get some info.

I see what you're saying. This test is meant to verify the specific code path where there's never a valid result returned from any node. All of the other tests in this file already test the error-free code paths.

I can add an additional test that expects a mix of some errors, some valid results.

zachlite

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @j82w, @knz, @maryliag, and @nkodali)

pkg/server/span_stats_test.go line 200 at r3 (raw file):

Previously, zachlite wrote…

I see what you're saying. This test is meant to verify the specific code path where there's never a valid result returned from any node. All of the other tests in this file already test the error-free code paths.

I can add an additional test that expects a mix of some errors, some valid results.

Done.

maryliag

one nit, otherwise

Reviewed 1 of 1 files at r4, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @j82w, @knz, @nkodali, and @zachlite)

pkg/server/span_stats_test.go line 290 at r4 (raw file):

				require.Equal(t, true, containsError(res.Errors, "node 4 timed out"))

				// There should not be any errors for node 2 or node 5.

you already tested that the length of error message is 3 and you checked for all of them, so there is no need to check these 2, specially because is not checking with right thing, there could have been a different error type on node 2 that is not dialing (same for node 5).

zachlite

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @j82w, @knz, @maryliag, and @nkodali)

pkg/server/span_stats_test.go line 290 at r4 (raw file):

Previously, maryliag (Marylia Gutierrez) wrote…

you already tested that the length of error message is 3 and you checked for all of them, so there is no need to check these 2, specially because is not checking with right thing, there could have been a different error type on node 2 that is not dialing (same for node 5).

Discussed offline. Keeping for documentation purposes.

zachlite · 2023-08-14T16:37:23Z

bors r+

j82w · 2023-08-14T17:32:18Z

bors r-

craig · 2023-08-14T17:32:20Z

Canceled.

yuzefovich · 2023-08-14T17:35:57Z

FYI the failure on Extended CI:

* ERROR: a panic has occurred!
* use of Span after Finish. Span: /cockroach.roachpb.Internal/Batch. Finish previously called at: <stack not captured. Set debugUseAfterFinish>
* (1) attached stack trace
*   -- stack trace:
*   | runtime.gopanic
*   | 	GOROOT/src/runtime/panic.go:890
*   | [...repeated from below...]
* Wraps: (2) assertion failure
* Wraps: (3) attached stack trace
*   -- stack trace:
*   | github.com/cockroachdb/cockroach/pkg/util/tracing.(*Span).detectUseAfterFinish
*   | 	github.com/cockroachdb/cockroach/pkg/util/tracing/span.go:182
*   | github.com/cockroachdb/cockroach/pkg/util/tracing.(*Span).Tracer
*   | 	github.com/cockroachdb/cockroach/pkg/util/tracing/span.go:225
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*replicaDecoder).createTracingSpans
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/replica_application_decoder.go:154
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*replicaDecoder).DecodeAndBind
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/replica_application_decoder.go:63
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver/apply.(*Task).Decode
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/apply/task.go:142
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*Replica).handleRaftReadyRaftMuLocked
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/replica_raft.go:878
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*Replica).handleRaftReady
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/replica_raft.go:728
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*Store).processReady
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/store_raft.go:689
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*raftSchedulerShard).worker
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/scheduler.go:418
*   | github.com/cockroachdb/cockroach/pkg/kv/kvserver.(*raftScheduler).Start.func2
*   | 	github.com/cockroachdb/cockroach/pkg/kv/kvserver/pkg/kv/kvserver/scheduler.go:321
*   | github.com/cockroachdb/cockroach/pkg/util/stop.(*Stopper).RunAsyncTaskEx.func2
*   | 	github.com/cockroachdb/cockroach/pkg/util/stop/stopper.go:484
*   | runtime.goexit
*   | 	src/runtime/asm_amd64.s:1594
* Wraps: (4) use of Span after Finish. Span: /cockroach.roachpb.Internal/Batch. Finish previously called at: <stack not captured. Set debugUseAfterFinish>
* Error types: (1) *withstack.withStack (2) *assert.withAssertionFailure (3) *withstack.withStack (4) *errutil.leafError
*
panic: use of Span after Finish. Span: /cockroach.roachpb.Internal/Batch. Finish previously called at: <stack not captured. Set debugUseAfterFinish> [recovered]
	panic: use of Span after Finish. Span: /cockroach.roachpb.Internal/Batch. Finish previously called at: <stack not captured. Set debugUseAfterFinish>

is a dup of #108534.

This commit adds improved fault tolerance to the span stats fan-out: 1. Errors encountered during the fan-out will not invalidate the entire request. Now, a span stats fan-out will always return a roachpb.SpanStatsResponse that has been updated by values from nodes that service their requests without error. In the extreme case where there's a failure encountered on every node, an empty response is returned. Errors that are encountered are logged, and then appended to the response in the newly added `Errors` field. 2. Nodes must service requests within the timeout passed to `iterateNodes`. For span stats, the value comes from a new cluster setting: 'server.span_stats.node.timeout', with a default value of 1 minute. Resolves cockroachdb#106097 Epic: none Release note (ops change): Span stats requests will return a partial result if the request encounters any errors. Errors that would have previously terminated the request are now included in the response.

zachlite · 2023-08-15T02:30:08Z

bors r+

craig · 2023-08-15T03:02:01Z

Build succeeded:

Bazel Essential CI (Cockroach)

blathers-crl · 2023-08-15T03:02:08Z

Encountered an error creating backports. Some common things that can go wrong:

The backport branch might have already existed.
There was a merge conflict.
The backport branch contained merge commits.

You might need to create your backport manually using the backport tool.

error creating merge commit from 5ddcdd1 to blathers/backport-release-23.1-108456: POST https://api.github.com/repos/cockroachdb/cockroach/merges: 409 Merge conflict []

you may need to manually resolve merge conflicts with the backport tool.

Backport to branch 23.1.x failed. See errors above.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

zachlite requested a review from a team as a code owner August 9, 2023 16:11

zachlite requested a review from a team August 9, 2023 16:11

zachlite requested a review from a team as a code owner August 9, 2023 16:11

zachlite requested review from a team and nkodali and removed request for a team August 9, 2023 16:11

zachlite added the backport-23.1.x Flags PRs that need to be backported to 23.1 label Aug 9, 2023

zachlite force-pushed the 230809.spanstats-fanout-fault-tolerance branch from cc968fd to bc57125 Compare August 9, 2023 16:28

j82w reviewed Aug 9, 2023

View reviewed changes

zachlite force-pushed the 230809.spanstats-fanout-fault-tolerance branch from bc57125 to 30c0898 Compare August 10, 2023 17:00

zachlite commented Aug 10, 2023

View reviewed changes

knz reviewed Aug 10, 2023

View reviewed changes

zachlite force-pushed the 230809.spanstats-fanout-fault-tolerance branch from 30c0898 to e8d6e16 Compare August 11, 2023 13:46

zachlite commented Aug 11, 2023

View reviewed changes

maryliag reviewed Aug 14, 2023

View reviewed changes

zachlite commented Aug 14, 2023

View reviewed changes

zachlite force-pushed the 230809.spanstats-fanout-fault-tolerance branch from e8d6e16 to 5713064 Compare August 14, 2023 15:57

zachlite commented Aug 14, 2023

View reviewed changes

maryliag approved these changes Aug 14, 2023

View reviewed changes

zachlite commented Aug 14, 2023

View reviewed changes

zachlite force-pushed the 230809.spanstats-fanout-fault-tolerance branch from 5713064 to b742181 Compare August 14, 2023 18:38

zachlite force-pushed the 230809.spanstats-fanout-fault-tolerance branch from b742181 to 5ddcdd1 Compare August 14, 2023 21:35

craig bot merged commit abf61bb into cockroachdb:master Aug 15, 2023

cockroach-teamcity mentioned this pull request Aug 15, 2023

PR #108456 - server: make the span stats fan-out more fault tolerant cockroachdb/docs#17665

Closed

zachlite mentioned this pull request Aug 18, 2023

release-23.1: server: make the span stats fan-out more fault tolerant #109008

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

server: make the span stats fan-out more fault tolerant #108456

server: make the span stats fan-out more fault tolerant #108456

zachlite commented Aug 9, 2023 •

edited

Loading

cockroach-teamcity commented Aug 9, 2023

j82w left a comment

j82w left a comment

zachlite left a comment

knz left a comment

zachlite left a comment

maryliag left a comment

zachlite left a comment

zachlite left a comment

maryliag left a comment

zachlite left a comment

zachlite commented Aug 14, 2023

j82w commented Aug 14, 2023

craig bot commented Aug 14, 2023

yuzefovich commented Aug 14, 2023

zachlite commented Aug 15, 2023

craig bot commented Aug 15, 2023

blathers-crl bot commented Aug 15, 2023

server: make the span stats fan-out more fault tolerant #108456

server: make the span stats fan-out more fault tolerant #108456

Conversation

zachlite commented Aug 9, 2023 • edited Loading

cockroach-teamcity commented Aug 9, 2023

j82w left a comment

Choose a reason for hiding this comment

j82w left a comment

Choose a reason for hiding this comment

zachlite left a comment

Choose a reason for hiding this comment

knz left a comment

Choose a reason for hiding this comment

zachlite left a comment

Choose a reason for hiding this comment

maryliag left a comment

Choose a reason for hiding this comment

zachlite left a comment

Choose a reason for hiding this comment

zachlite left a comment

Choose a reason for hiding this comment

maryliag left a comment

Choose a reason for hiding this comment

zachlite left a comment

Choose a reason for hiding this comment

zachlite commented Aug 14, 2023

j82w commented Aug 14, 2023

craig bot commented Aug 14, 2023

yuzefovich commented Aug 14, 2023

zachlite commented Aug 15, 2023

craig bot commented Aug 15, 2023

blathers-crl bot commented Aug 15, 2023

zachlite commented Aug 9, 2023 •

edited

Loading