Skip to content

Antalya 26.3 - Port CI fixes - #2621

Open
CarlosFelipeOR wants to merge 8 commits into
antalya-26.3from
ci/fixes-antalya-26.3.53
Open

CarlosFelipeOR wants to merge 8 commits into
antalya-26.3from
ci/fixes-antalya-26.3.53

Conversation

@CarlosFelipeOR

@CarlosFelipeOR CarlosFelipeOR commented Oct 20, 2026 •

Copy link
Copy Markdown
Collaborator

Ports to antalya-26.3 CI fixes that already landed on other branches.

From stable-26.3:

From antalya-26.6:

From upstream:

Changelog category (leave one):

  • CI Fix or Improvement (changelog entry is not required)

CI/CD Options

Exclude tests:

  • Fast test
  • Integration Tests
  • Stateless tests
  • Stateful tests
  • Unit tests
  • Performance tests
  • Aarch64 tests
  • All with ASAN
  • All with TSAN
  • All with MSAN
  • All with UBSAN
  • All with Coverage
  • All Regression
  • Disable CI Cache

Regression jobs to run:

  • Fast suites (mostly <1h)
  • Aggregate Functions (2h)
  • Alter (1.5h)
  • Benchmark (30m)
  • CAS (content-addressed storage; Antalya only)
  • ClickHouse Keeper (1h)
  • Iceberg (2h)
  • LDAP (1h)
  • OAuth (5m)
  • Parquet (1.5h)
  • RBAC (1.5h)
  • SSL Server (1h)
  • S3 (2h)
  • S3 Export (2h)
  • Swarms (30m)
  • Tiered Storage (2h)

alexey-milovidov and others added 5 commits October 20, 2026 10:17
…ilesystem-cache-path

Fix ClickBench job: allow `/dev/shm/clickhouse/` as filesystem cache path

(cherry picked from commit 48ea778)
(cherry picked from commit 7d710d7)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
The module gained with_kafka=True from the ClickHouse#117737 backport, but 26.3
lacks the arm64 Kafka image override, so zookeeper:3.4.9 never starts
under emulation. Every other Kafka test in the suite already skips on ARM.

Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
(cherry picked from commit 5dd45e2)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit 4c2d531)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
(cherry picked from commit 3c8e750)
(cherry picked from commit 4016b19)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
Mark `05082_kill_and_system_on_cluster_access_types` with
`no-encrypted-storage`: this configuration disables transactions, so
`system.transactions` is absent and the successful `KILL TRANSACTION`
checks fail with `UNKNOWN_TABLE`.

CI report:
https://altinity-build-artifacts.s3.amazonaws.com/json.html?PR=2589&sha=25cd28d33e10b51887ff15b464c2adab26e3fe80&name_0=PR&name_1=Stateless+tests+%28amd_debug%2C+distributed+plan%2C+s3+storage%2C+parallel%29&name_2=Tests

(cherry picked from commit 4980aee)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
@CarlosFelipeOR CarlosFelipeOR added antalya cicd Improvements and fixes to the CICD process antalya-26.3 antalya-26.3.53.20001 labels Oct 20, 2026
@github-actions

github-actions Bot commented Oct 20, 2026 •

Copy link
Copy Markdown

Workflow [PR], commit [bc14708]

CarlosFelipeOR and others added 2 commits October 20, 2026 14:44
…e database disk prefix is not shared

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
(cherry picked from commit cd1ec63)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
test_concurrent_reads runs 25 concurrent SELECT * queries against one
accumulating Iceberg table while 15 Spark writers commit in parallel, for
10 rounds. By default each SELECT fans out to ~max_threads (number of CPU
cores) internal read threads, so 25 simultaneous readers request hundreds
of threads at once. On a loaded CI shard this oversubscribes the global
thread pool: a single SELECT either gets starved past the 300s client
receive_timeout (Return code 159, seen on the asan/db-disk/old-analyzer 5/6
shard for PRs ClickHouse#100651 and ClickHouse#108049 and on master) or the server rejects the
query outright with CANNOT_SCHEDULE_TASK (Code 439, threads=1000, jobs=1000).

Pin max_threads=4 on the concurrent SELECT so the readers' aggregate thread
demand stays bounded. This does not change what the test verifies: all 25
readers still race all 15 writers across every round and the exact row-count
assertion is unchanged. Only the per-query intra-read parallelism is capped,
which is irrelevant to concurrent-read snapshot consistency.

Locally reproduced both directions on a debug build: pristine test fails at
round 2 with CANNOT_SCHEDULE_TASK; with the cap it passes all 10 rounds
(~150s).

(cherry picked from commit 0ee2f26)
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>
@CarlosFelipeOR CarlosFelipeOR changed the title Antalya 26.3 - port stable-26.3 CI fixes Antalya 26.3 - Port CI fixes Oct 20, 2026
Grype Scan reports High CVEs in libssl3/libcrypto3 and zlib in both
alpine images. Running `apk upgrade` at build time pulls the fixed
packages, as stable-26.3 already does since #2590.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Signed-off-by: CarlosFelipeOR <carlosfelipeor@gmail.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

antalya antalya-26.3 antalya-26.3.53.20001 cicd Improvements and fixes to the CICD process

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants