Skip to content

Antalya 26.3 Backport of #131001 - Reuse the GCM cipher context between rows in encrypt/decrypt - #2898

Merged
zvonand merged 1 commit into
antalya-26.3from
backports/antalya-26.3/131001
Nov 18, 2026
Merged

zvonand merged 1 commit into
antalya-26.3from
backports/antalya-26.3/131001

Conversation

@blau-ai

@blau-ai blau-ai commented Nov 18, 2026

Copy link
Copy Markdown
Collaborator

Changelog category (leave one):

  • Performance Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Improve performance of the encrypt, decrypt and tryDecrypt functions in AES-GCM modes by up to 27x with many threads. (ClickHouse#131001 by @mkmkme)

Documentation entry for user-facing changes

...

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)

Reuse the GCM cipher context between rows in encrypt/decrypt
@github-actions

Copy link
Copy Markdown

Workflow [PR], commit [30af90b]

@zvonand

zvonand commented Nov 18, 2026

Copy link
Copy Markdown
Member

@blau-ai

@blau-ai

blau-ai commented Nov 18, 2026

Copy link
Copy Markdown
Collaborator Author

CI triage for 30af90b

Verdict: none of the failures come from this PR. 0 caused by the PR, 3 pre-existing CI-environment problems, 1 new scan result that is not about the PR code, and 1 expected DCO failure.

The PR only touches src/Functions/FunctionsAES.h plus an encrypt/decrypt stateless test and a perf test. Every check that exercises that code passed: all builds that ran, Fast test, stateless tests (asan / debug / ubsan, including the s3 and distributed-plan variants), unit tests (asan, ubsan), stress tests, AST fuzzer and BuzzHouse. The tsan/msan/arm jobs were skipped by the CI config for this run, not by failures.

Check Failing tests Classification
Integration tests (amd_binary, 2/5) and (amd_asan, db disk, old analyzer, 3/8) test_storage_s3/test_sts.py ×3 Pre-existing, CI environment
Integration tests (amd_binary, 4/5) and (amd_asan, …, 5/8) test_database_glue/test.py ×18 Pre-existing, CI environment
Integration tests (amd_asan, …, 5/8) test_database_hms/test.py::test_hive_catalog_url_parsing Infra: Docker image pull hung
SQLLogic test — Pre-existing timeout
Grype scans (keeper and server-alpine images) 4 High CVEs Not from the PR's code; new CVE entries, see below
DCO — Expected for backport merge commits

1. test_sts.py and test_database_glue (21 tests in 4 shards): pre-existing

Every one of these fails during cluster.start(), before any query runs. The mock STS container cannot start:

E   Exception: Timed out while waiting for instance `sts.us-east-1.amazonaws.com' with ip address 172.16.3.8 to start.
E   bash: line 1: /usr/bin/clickhouse: cannot execute: required file not found

All 21 failures end with that same line. The same tests fail with the same error on PRs that have nothing to do with this code:

"required file not found" for an existing binary path usually means the binary is there but its ELF interpreter or loader is missing inside the mock container image. So the problem is the STS mock image or how the binary is mounted into it, not ClickHouse code. Next step: nothing to do on this PR. Someone should fix the STS mock image on the antalya-26.3 CI.

2. test_database_hms::test_hive_catalog_url_parsing: infra

This test also timed out in cluster.start(), while pulling Docker images:

File: helpers/cluster.py:3513 - in start
    retry(...)(run_and_check, images_pull_cmd, timeout=600)
E   Failed: Timeout (>900.0s) from pytest-timeout.

It never reached the HMS/URL-parsing logic, which this PR does not touch anyway. Safe to re-run.

3. SQLLogic test: pre-existing

The report says only Timeout. #2871 hit the same Timeout after 3h1m. This is not related to the PR.

4. Grype scans: 4 High CVEs in the Alpine-based images

clickhouse-keeper and clickhouse-server-alpine both report the same set: CVE-2026-95692, CVE-2026-95691, CVE-2026-88823 and CVE-2026-107496, all nvd:cpe, High. The Debian-based clickhouse-server scan passed. On #2871, one day earlier with the same 26.3.64 version, all three scans reported 0 high/critical findings. This PR changes a C++ header and tests, not the image contents, so most likely new advisories were published that match a package in the Alpine base image. The report doesn't name the affected package, so this is an inference. Next step: this belongs to the image/base-layer owners, not to this PR. It will probably show up on every antalya-26.3 PR from now on.

5. DCO: expected

The head commit is upstream's merge commit (Merge pull request #131001 …), which has no Signed-off-by. DCO fails the same way on other backport PRs (e.g. #2871). Ignore it, or set DCO to pass manually as usual.


Bottom line: the backport looks good from CI's point of view, and no changes to the PR are needed. You can re-run the HMS shard if you want a cleaner board, but the STS/Glue, SQLLogic and Grype failures will stay red until they are fixed on the base branch or the CI images.

@zvonand
zvonand merged commit ee91964 into antalya-26.3 Nov 18, 2026
400 of 425 checks passed
@zvonand zvonand added the verified Approved for release label Nov 18, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants