kvstorage: add WAG replay logic by miraradeva · Pull Request #167029 · cockroachdb/cockroach

miraradeva · 2026-03-30T14:44:49Z

Add replay infrastructure for applying WAG (Write-Ahead Graph) nodes to
the state machine during store startup. The replay loop iterates over
the WAG in the log engine and applies any unapplied mutations.

The core decision logic is split into layers:

canApply: per-event decision based on the event's address vs. the
range's persistedRangeState (replica ID, tombstone, applied index).
raftCatchUp: determines the raft index a replica must be caught up to
before a WAG node can be applied. Varies by event type (e.g. splits
catch up to Index-1, destroys to Index, creates need no catch-up).
canApplyWAGNode: node-level wrapper that checks all events agree on
whether to apply (returns error if not) and collects per-range raft
catch-up targets.
ReplayWAG: iterates the WAG log and applies unapplied mutations.

Resolves: #167278
Epic: CRDB-55218
Release note: None

trunk-io · 2026-03-30T14:44:54Z

😎 Merged successfully - details.

cockroach-teamcity · 2026-03-30T14:45:05Z

This change is

cockroach-teamcity · 2026-04-01T15:10:15Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.31m ±2%	10.21m ±5%	~	p=0.683 n=15
⚪ allocs/op	8.122k ±1%	8.116k ±0%	~	p=0.935 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/b4c5921/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/b4c5921fb83adf189e9bc2d62e77ae56d90f838e/bin/pkg_sql_tests benchdiff/b4c5921/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/b4c5921/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/6354858/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/6354858a402243510fa3abbbdc2a3c1a0a7c659e/bin/pkg_sql_tests benchdiff/6354858/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/6354858/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=6354858 --new=b4c5921 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.085m ±1%	3.091m ±1%	~	p=0.595 n=15
⚪ allocs/op	2.106k ±0%	2.106k ±0%	~	p=0.913 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/b4c5921/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/b4c5921fb83adf189e9bc2d62e77ae56d90f838e/bin/pkg_sql_tests benchdiff/b4c5921/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/b4c5921/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/6354858/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/6354858a402243510fa3abbbdc2a3c1a0a7c659e/bin/pkg_sql_tests benchdiff/6354858/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/6354858/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=6354858 --new=b4c5921 --memprofile ./pkg/sql/tests

🔴 Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
🔴 sec/op	2.954m ±1%	2.993m ±2%	+1.34%	p=0.002 n=15
⚪ allocs/op	4.207k ±0%	4.212k ±0%	+0.12%	p=0.016 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/b4c5921/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/b4c5921fb83adf189e9bc2d62e77ae56d90f838e/bin/pkg_sql_tests benchdiff/b4c5921/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/b4c5921/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/6354858/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/6354858a402243510fa3abbbdc2a3c1a0a7c659e/bin/pkg_sql_tests benchdiff/6354858/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/6354858/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=6354858 --new=b4c5921 --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/b4c5921fb83adf189e9bc2d62e77ae56d90f838e/23854081926-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/6354858a402243510fa3abbbdc2a3c1a0a7c659e/23854081926-1/\* old/

built with commit: b4c5921fb83adf189e9bc2d62e77ae56d90f838e

cockroach-teamcity · 2026-04-01T16:49:30Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.12m ±3%	10.24m ±3%	~	p=0.512 n=15
⚪ allocs/op	8.114k ±0%	8.090k ±1%	~	p=0.174 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/2b66fae/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/2b66faeadd990d9d93658be62b6cd467f1156c4a/bin/pkg_sql_tests benchdiff/2b66fae/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/2b66fae/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=c2e8221 --new=2b66fae --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.124m ±2%	3.107m ±2%	~	p=0.967 n=15
⚪ allocs/op	2.106k ±0%	2.106k ±0%	~	p=0.930 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/2b66fae/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/2b66faeadd990d9d93658be62b6cd467f1156c4a/bin/pkg_sql_tests benchdiff/2b66fae/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/2b66fae/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=c2e8221 --new=2b66fae --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.003m ±1%	2.990m ±1%	~	p=0.033 n=15
⚪ allocs/op	4.211k ±0%	4.209k ±0%	~	p=0.390 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/2b66fae/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/2b66faeadd990d9d93658be62b6cd467f1156c4a/bin/pkg_sql_tests benchdiff/2b66fae/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/2b66fae/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=c2e8221 --new=2b66fae --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/2b66faeadd990d9d93658be62b6cd467f1156c4a/23859203948-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c2e82219150d1858a5542db18805ce06c43bf317/23859203948-1/\* old/

built with commit: 2b66faeadd990d9d93658be62b6cd467f1156c4a

cockroach-teamcity · 2026-04-01T17:36:40Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.47m ±2%	10.58m ±2%	~	p=0.539 n=15
⚪ allocs/op	8.120k ±1%	8.119k ±1%	~	p=0.443 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/581d02c/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/581d02cfb4d789bce0558e29ae7962e6265d6db1/bin/pkg_sql_tests benchdiff/581d02c/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/581d02c/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=c2e8221 --new=581d02c --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.121m ±1%	3.130m ±1%	~	p=0.412 n=15
⚪ allocs/op	2.106k ±0%	2.106k ±0%	~	p=0.931 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/581d02c/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/581d02cfb4d789bce0558e29ae7962e6265d6db1/bin/pkg_sql_tests benchdiff/581d02c/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/581d02c/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=c2e8221 --new=581d02c --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	2.885m ±1%	2.902m ±1%	~	p=0.161 n=15
⚪ allocs/op	4.204k ±0%	4.203k ±0%	~	p=0.645 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/581d02c/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/581d02cfb4d789bce0558e29ae7962e6265d6db1/bin/pkg_sql_tests benchdiff/581d02c/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/581d02c/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=c2e8221 --new=581d02c --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/581d02cfb4d789bce0558e29ae7962e6265d6db1/23861267661-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c2e82219150d1858a5542db18805ce06c43bf317/23861267661-1/\* old/

built with commit: 581d02cfb4d789bce0558e29ae7962e6265d6db1

pav-kv

First pass, didn't look at tests yet. Looks great.

pav-kv · 2026-04-01T19:44:14Z

CC @iskettaneh to get familiar with the WAG encoding etc. The truncation logic will have some overlaps.

cockroach-teamcity · 2026-04-02T13:47:11Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	11.01m ±5%	10.94m ±5%	~	p=0.367 n=15
⚪ allocs/op	8.147k ±3%	8.129k ±2%	~	p=0.512 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/9450917/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/94509177f462ac9b4b8083fc43a3433172629b89/bin/pkg_sql_tests benchdiff/9450917/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/9450917/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=c2e8221 --new=9450917 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.296m ±2%	3.332m ±2%	~	p=0.161 n=15
⚪ allocs/op	2.107k ±0%	2.107k ±0%	~	p=0.063 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/9450917/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/94509177f462ac9b4b8083fc43a3433172629b89/bin/pkg_sql_tests benchdiff/9450917/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/9450917/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=c2e8221 --new=9450917 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.212m ±3%	3.176m ±3%	~	p=0.137 n=15
⚪ allocs/op	4.225k ±0%	4.226k ±0%	~	p=0.629 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/9450917/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/94509177f462ac9b4b8083fc43a3433172629b89/bin/pkg_sql_tests benchdiff/9450917/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/9450917/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=c2e8221 --new=9450917 --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/94509177f462ac9b4b8083fc43a3433172629b89/23902650591-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c2e82219150d1858a5542db18805ce06c43bf317/23902650591-1/\* old/

built with commit: 94509177f462ac9b4b8083fc43a3433172629b89

cockroach-teamcity · 2026-04-03T15:43:40Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.00m ±1%	10.02m ±1%	~	p=0.345 n=15
⚪ allocs/op	8.079k ±0%	8.125k ±1%	~	p=0.044 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/c668ce6/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c668ce690c03ff0d77dfd608c8e74d3183dea630/bin/pkg_sql_tests benchdiff/c668ce6/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c668ce6/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=c2e8221 --new=c668ce6 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.064m ±1%	3.074m ±1%	~	p=0.367 n=15
⚪ allocs/op	2.106k ±0%	2.106k ±1%	~	p=0.673 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/c668ce6/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c668ce690c03ff0d77dfd608c8e74d3183dea630/bin/pkg_sql_tests benchdiff/c668ce6/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c668ce6/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=c2e8221 --new=c668ce6 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	2.891m ±1%	2.892m ±0%	~	p=0.250 n=15
⚪ allocs/op	4.201k ±0%	4.204k ±0%	~	p=0.132 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/c668ce6/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c668ce690c03ff0d77dfd608c8e74d3183dea630/bin/pkg_sql_tests benchdiff/c668ce6/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c668ce6/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=c2e8221 --new=c668ce6 --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c668ce690c03ff0d77dfd608c8e74d3183dea630/23951375097-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c2e82219150d1858a5542db18805ce06c43bf317/23951375097-1/\* old/

built with commit: c668ce690c03ff0d77dfd608c8e74d3183dea630

cockroach-teamcity · 2026-04-06T14:30:33Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.57m ±2%	10.56m ±1%	~	p=0.683 n=15
⚪ allocs/op	8.118k ±1%	8.113k ±1%	~	p=0.814 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/a0cde21/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/a0cde2154e7ee8a0924b228aaace8ddaf88bfa54/bin/pkg_sql_tests benchdiff/a0cde21/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/a0cde21/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=c2e8221 --new=a0cde21 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.099m ±1%	3.103m ±1%	~	p=0.902 n=15
⚪ allocs/op	2.106k ±0%	2.107k ±0%	~	p=0.466 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/a0cde21/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/a0cde2154e7ee8a0924b228aaace8ddaf88bfa54/bin/pkg_sql_tests benchdiff/a0cde21/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/a0cde21/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=c2e8221 --new=a0cde21 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	2.925m ±2%	2.923m ±2%	~	p=0.461 n=15
⚪ allocs/op	4.207k ±0%	4.207k ±0%	~	p=0.798 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/a0cde21/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/a0cde2154e7ee8a0924b228aaace8ddaf88bfa54/bin/pkg_sql_tests benchdiff/a0cde21/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/a0cde21/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=c2e8221 --new=a0cde21 --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/a0cde2154e7ee8a0924b228aaace8ddaf88bfa54/24035030765-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c2e82219150d1858a5542db18805ce06c43bf317/24035030765-1/\* old/

built with commit: a0cde2154e7ee8a0924b228aaace8ddaf88bfa54

pav-kv

LGTM % minor suggestions.

Add replay infrastructure for applying WAG (Write-Ahead Graph) nodes to the state machine during store startup. The replay loop iterates over the WAG in the log engine and applies any unapplied mutations. The core decision logic is split into layers: - canApply: per-event decision based on the event's address vs. the range's persistedRangeState (replica ID, tombstone, applied index). - raftCatchUp: determines the raft index a replica must be caught up to before a WAG node can be applied. Varies by event type (e.g. splits catch up to Index-1, destroys to Index, creates need no catch-up). - canApplyWAGNode: node-level wrapper that checks all events agree on whether to apply (returns error if not) and collects per-range raft catch-up targets. - ReplayWAG: iterates the WAG log and applies unapplied mutations. Resolves: cockroachdb#167278 Epic: CRDB-55218 Release note: None Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

cockroach-teamcity · 2026-04-08T17:14:26Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.08m ±2%	10.28m ±3%	~	p=0.056 n=15
⚪ allocs/op	8.083k ±1%	8.101k ±1%	~	p=0.290 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/b27ce04/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/b27ce04e9261df6408d0083d2f71014c81e05a18/bin/pkg_sql_tests benchdiff/b27ce04/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/b27ce04/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=c2e8221 --new=b27ce04 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.187m ±2%	3.190m ±2%	~	p=0.935 n=15
⚪ allocs/op	2.106k ±0%	2.106k ±0%	~	p=0.460 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/b27ce04/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/b27ce04e9261df6408d0083d2f71014c81e05a18/bin/pkg_sql_tests benchdiff/b27ce04/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/b27ce04/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=c2e8221 --new=b27ce04 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.037m ±1%	3.030m ±1%	~	p=0.713 n=15
⚪ allocs/op	4.216k ±0%	4.213k ±0%	~	p=0.062 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/b27ce04/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/b27ce04e9261df6408d0083d2f71014c81e05a18/bin/pkg_sql_tests benchdiff/b27ce04/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/b27ce04/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/c2e8221/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/c2e82219150d1858a5542db18805ce06c43bf317/bin/pkg_sql_tests benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/c2e8221/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=c2e8221 --new=b27ce04 --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/b27ce04e9261df6408d0083d2f71014c81e05a18/24147510321-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/c2e82219150d1858a5542db18805ce06c43bf317/24147510321-1/\* old/

built with commit: b27ce04e9261df6408d0083d2f71014c81e05a18

cockroach-teamcity · 2026-04-08T19:59:08Z

⚪ Sysbench [SQL, 3node, oltp_read_write]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	10.24m ±1%	10.22m ±1%	~	p=0.713 n=15
⚪ allocs/op	8.087k ±1%	8.107k ±0%	~	p=0.943 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/7696e69/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/7696e690fb5f9167e6cfbf2984f38dd864d4ec53/bin/pkg_sql_tests benchdiff/7696e69/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/7696e69/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/8922fce/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/8922fce6a04bb5a8433d7c1daf3ff4770b8739b1/bin/pkg_sql_tests benchdiff/8922fce/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/8922fce/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/SQL/3node/oltp_read_write$ --old=8922fce --new=7696e69 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_read_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	3.149m ±2%	3.125m ±2%	~	p=0.161 n=15
⚪ allocs/op	2.106k ±0%	2.106k ±0%	~	p=0.154 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/7696e69/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/7696e690fb5f9167e6cfbf2984f38dd864d4ec53/bin/pkg_sql_tests benchdiff/7696e69/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/7696e69/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/8922fce/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/8922fce6a04bb5a8433d7c1daf3ff4770b8739b1/bin/pkg_sql_tests benchdiff/8922fce/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/8922fce/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_read_only$ --old=8922fce --new=7696e69 --memprofile ./pkg/sql/tests

⚪ Sysbench [KV, 3node, oltp_write_only]

Metric	Old Commit	New Commit	Delta	Note
⚪ sec/op	2.949m ±0%	2.940m ±1%	~	p=0.305 n=15
⚪ allocs/op	4.208k ±0%	4.207k ±0%	~	p=0.307 n=15

Reproduce

benchdiff binaries:

mkdir -p benchdiff/7696e69/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/7696e690fb5f9167e6cfbf2984f38dd864d4ec53/bin/pkg_sql_tests benchdiff/7696e69/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/7696e69/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
mkdir -p benchdiff/8922fce/bin/1058449141
gcloud storage cp gs://cockroach-microbench-ci/builds/8922fce6a04bb5a8433d7c1daf3ff4770b8739b1/bin/pkg_sql_tests benchdiff/8922fce/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests
chmod +x benchdiff/8922fce/bin/1058449141/cockroachdb_cockroach_pkg_sql_tests

benchdiff command:

# NB: for best (most stable) results, also add a suitable `--benchtime` that
# results in ~1s to ~5s of benchmark runs. For example, if ops average ~3ms, a
# benchtime of `1000x` is appropriate.
#
# Some benchmarks (in particular BenchmarkSysbench) output additional memory
# profiles covering only the execution (excluding the setup/teardown) - those
# should be preferred for analysis since they more closely correspond to what's
# reported as B/op and alloc/op.
benchdiff --run=^BenchmarkSysbench/KV/3node/oltp_write_only$ --old=8922fce --new=7696e69 --memprofile ./pkg/sql/tests

Artifacts

download:

mkdir -p new
gcloud storage cp gs://cockroach-microbench-ci/artifacts/7696e690fb5f9167e6cfbf2984f38dd864d4ec53/24154608575-1/\* new/
mkdir -p old
gcloud storage cp gs://cockroach-microbench-ci/artifacts/8922fce6a04bb5a8433d7c1daf3ff4770b8739b1/24154608575-1/\* old/

built with commit: 7696e690fb5f9167e6cfbf2984f38dd864d4ec53

miraradeva requested a review from a team as a code owner March 30, 2026 14:44

miraradeva requested a review from pav-kv March 30, 2026 14:45

miraradeva force-pushed the mira/wag-replay branch from 6e39d71 to b4c5921 Compare April 1, 2026 14:34

cockroach-teamcity added the X-perf-check Microbenchmarks CI: Added to a PR if a performance regression is detected and should be checked label Apr 1, 2026

miraradeva force-pushed the mira/wag-replay branch 2 times, most recently from 15c03d1 to 2b66fae Compare April 1, 2026 16:26

miraradeva force-pushed the mira/wag-replay branch from 2b66fae to 581d02c Compare April 1, 2026 17:15

pav-kv reviewed Apr 1, 2026

View reviewed changes

miraradeva force-pushed the mira/wag-replay branch from 581d02c to 9450917 Compare April 2, 2026 13:24

pav-kv reviewed Apr 2, 2026

View reviewed changes

miraradeva force-pushed the mira/wag-replay branch from 9450917 to a976658 Compare April 3, 2026 15:19

miraradeva changed the title ~~kvstorage/wag: add WAG replay logic~~ kvstorage: add WAG replay logic Apr 3, 2026

miraradeva force-pushed the mira/wag-replay branch from a976658 to c668ce6 Compare April 3, 2026 15:21

miraradeva force-pushed the mira/wag-replay branch from c668ce6 to a0cde21 Compare April 6, 2026 14:07

pav-kv approved these changes Apr 7, 2026

View reviewed changes

miraradeva force-pushed the mira/wag-replay branch from a0cde21 to b27ce04 Compare April 8, 2026 16:48

miraradeva force-pushed the mira/wag-replay branch from b27ce04 to 7696e69 Compare April 8, 2026 19:35

pav-kv approved these changes Apr 8, 2026

View reviewed changes

trunk-io Bot merged commit 8de24a1 into cockroachdb:master Apr 8, 2026
30 checks passed

celeste-cockroachdb Bot added the target-release-26.3.0 label Apr 8, 2026

Conversation

miraradeva commented Mar 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

trunk-io Bot commented Mar 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cockroach-teamcity commented Mar 30, 2026

Uh oh!

cockroach-teamcity commented Apr 1, 2026

Uh oh!

cockroach-teamcity commented Apr 1, 2026

Uh oh!

cockroach-teamcity commented Apr 1, 2026

Uh oh!

pav-kv left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

pav-kv commented Apr 1, 2026

Uh oh!

cockroach-teamcity commented Apr 2, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cockroach-teamcity commented Apr 3, 2026

Uh oh!

cockroach-teamcity commented Apr 6, 2026

Uh oh!

pav-kv left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

cockroach-teamcity commented Apr 8, 2026

Uh oh!

cockroach-teamcity commented Apr 8, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

miraradeva commented Mar 30, 2026 •

edited

Loading

trunk-io Bot commented Mar 30, 2026 •

edited

Loading