-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
storage: improve mvcc benchmarks under -short #115810
Conversation
ca53a69
to
0190fb0
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @jbowens and @sumeerbhola)
pkg/storage/bench_pebble_test.go
line 330 at r1 (raw file):
var testCases []testCase for _, batch := range []bool{false, true} { for _, numVersions := range []int{1, 10, 100} {
Note: here I am restoring the variants which were removed in #99726
29e0df0
to
e933c3b
Compare
Added a commit that fixes #115811. |
e933c3b
to
3d4322a
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Reviewed 4 of 4 files at r4, 2 of 2 files at r5, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @nvanbenschoten, @RaduBerinde, and @sumeerbhola)
-- commits
line 5 at r4:
nit: "skipped when"
pkg/storage/bench_pebble_test.go
line 330 at r1 (raw file):
Previously, RaduBerinde wrote…
Note: here I am restoring the variants which were removed in #99726
Oops, thanks
Many of the MVCC benchmarks have too many variants and are completely skipped when using the `-short` flag (which is used for the weekly microbenchmarks job). Conversely, other benchmarks are not skipped but still have quite a few variants. In this commit we change relevant tests to run a small set of configurations under `-short`. We also fix a few smaller issues, for example we remove some bogus empty acquire lock benchmarks with `heldOtherTxn=true/heldSameTxn=true`. Fixes cockroachdb#115794 Release note: None
This benchmark has batching variants but the batch size is `b.N` which can hit the 4GB batch limit. Given that we already have a separate `MVCCBatchPut` which has reasonable batch sizes, we are removing the batched variants of `BenchmarkMVCCPut`. Fixes: cockroachdb#115811 Release note: None
3d4322a
to
dd489c3
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
TFTR!
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale) (waiting on @jbowens, @nvanbenschoten, and @sumeerbhola)
Previously, jbowens (Jackson Owens) wrote…
nit: "skipped when"
Haha, I like to sip my benchmarks. Fixed.
bors r+ |
Build succeeded: |
Encountered an error creating backports. Some common things that can go wrong:
You might need to create your backport manually using the backport tool. error creating merge commit from 5ec3a5a to blathers/backport-release-23.2-115810: 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 23.2.x failed. See errors above. error creating merge commit from 5ec3a5a to blathers/backport-release-23.2.0-rc-115810: 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 23.2.0-rc failed. See errors above. 🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf. |
storage: improve mvcc benchmarks under -short
Many of the MVCC benchmarks have too many variants and are completely
sipped when using the
-short
flag (which is used for the weeklymicrobenchmarks job). Conversely, other benchmarks are not skipped but
still have quite a few variants.
In this commit we change relevant tests to run a small set of
configurations under
-short
. We also fix a few smaller issues, forexample we remove some bogus empty acquire lock benchmarks with
heldOtherTxn=true/heldSameTxn=true
.Fixes #115794
Release note: None
storage: remove batch=true variants of BenchmarkMVCCPut
This benchmark has batching variants but the batch size is
b.N
whichcan hit the 4GB batch limit.
Given that we already have a separate
MVCCBatchPut
which hasreasonable batch sizes, we are removing the batched variants of
BenchmarkMVCCPut
.Fixes: #115811
Release note: None