[zvonand]

Changelog category (leave one):

• Bug Fix

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

Fixed ALTER MODIFY COLUMN failing when converting a Nullable column to a different type with a DEFAULT (e.g. Nullable(UInt8) → String,LowCardinality(String), or Array(UInt8)). (ClickHouse#102156 by @il9ue)

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)

[github-actions]

Workflow [PR], commit [ede57c1]

[zvonand]

@blau-ai

[blau-ai]

CI triage for ede57c1

TL;DR: none of the failures are caused by this PR. There are 3 red checks (the PR aggregate just mirrors them):

Check	Failing tests	Verdict
Stateless tests (amd_debug, sequential)	03271_s3_table_function_asterisk_glob, 02789_filesystem_cache_alignment (+2 that passed on retry)	Already failing on the base branch
Stateless tests (amd_debug, distributed plan, s3 storage, sequential)	03271_s3_table_function_asterisk_glob, 02789_filesystem_cache_alignment	Already failing on the base branch
DCO	n/a	Metadata only: the commit has no Signed-off-by

The PR changes only src/Interpreters/inplaceBlockConversions.cpp and adds the test 03575_modify_column_null_to_default_cross_type. All parallel suites (asan/debug/ubsan, 0 failures) passed, and so did the ubsan and asan-db-disk sequential suites. None of the failing tests touch ALTER MODIFY COLUMN or Nullable conversion.

---

1. 03271_s3_table_function_asterisk_glob (both debug sequential suites, failed again on retry)

Error on processing query: Timeout exceeded while receiving data from server. Waited for 300 seconds, timeout is 300 seconds.
(query: SELECT * FROM s3('https://clickhouse-public-datasets.s3.amazonaws.com/wikistat/original/*', NOSIGN) LIMIT 1 FORMAT Null;)

• The query that hangs lists a large external public S3 bucket (wikistat/original/*), so the timeout comes from the runner's network or the debug build's speed, not from ALTER code.
• Evidence it's already broken: Antalya 26.3: Bump to 26.3.60 #2707 (the version bump to 26.3.60, no code changes) fails the same test in the same two suites (report).

2. 02789_filesystem_cache_alignment (both debug sequential suites, failed again on retry)

02789_filesystem_cache_alignment: [ FAIL ] 600.04 sec.
Reason: Timeout! Processes left in process group ...  (stuck in the SELECT over system.remote_data_paths / filesystem cache segments)

• This is a filesystem-cache test, and it hit the 600 s test timeout.
• Evidence it's already broken: Antalya 26.3: Bump to 26.3.60 #2707 fails it the same way in both suites.
• The CI "known broken" rule for timeouts ("Random timeout with sanitizer") only covers tsan/asan/msan/ubsan builds. That's why these timeouts aren't hidden on amd_debug (see broken_tests_handler.log).

3. Tests that failed once and passed on retry (amd_debug, sequential)

• 04306_delta_lake_merge_missing_table_state: got \N instead of 1/2 on the first run (a failpoint test for DeltaLake). Passed on retry. Unrelated to this PR.
• 02346_text_index_hits: TOO_SLOW (estimated 1577 s, maximum 600) while inserting test.hits, which is the debug build being slow. Passed on retry.

4. DCO

The only commit, ede57c1a ("Merge pull request ClickHouse#102156 from il9ue/…", the upstream merge commit cherry-picked as-is), has no Signed-off-by: line. Other 26.3 backports have the same problem (e.g. #2697 is also red on DCO).
Fix options:

• A maintainer marks DCO as passed from the check's details page. This is the usual approach for upstream backports.
• Or someone re-signs the commit (git commit --amend -s) and force-pushes. I can't do that because force-push is outside my permissions.

Adding a new signed commit on top won't help, because DCO checks every commit in the PR.

Suggested next steps

• No code changes needed in this PR.
• If you want the sequential suites green, re-run them. 03271/02789 will probably fail again until they're fixed on the antalya-26.3 base branch, since they currently fail on unrelated PRs too. A follow-up could add them to tests/broken_tests.yaml for amd_debug, or make 03271 stop depending on listing the public wikistat bucket.
• Fix DCO as above.

_{Evidence: result_pr.json, job.log and broken_tests_handler.log for this PR's head, compared with the reports for #2707, #2714, #2697 and #2684. I did not run anything locally.}