You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 954fde3
Browse filesBrowse the repository at this point in the historyBrowse files
The daily external regression pipeline for branch-4.0 fails 10 cases. Eight
of them turned red at the same build and stayed red; the other two are
long-standing.
1. Kerberos: the lightweight Kerberos environment creates every principal
with `addprinc -randkey`, so `/keytabs/*.keytab` holds a different key on
every container start. Deployments that run Doris on hosts other than the
Docker host provision `/keytabs` out of band and cannot follow that
rotation, so every Kerberos catalog fails with
`GeneralSecurityException: Checksum failed`. The old image-based
environment shipped fixed keys, which is why this only appeared after the
switch. Derive the keys from a fixed password and export them with
`ktadd -norandkey` instead: kdc.conf already pins a single enctype and the
salt is fixed, so the keytab is now reproducible across rebuilds.
2. Paimon HMS: `create_paimon_tables.hql` only runs from
`init-hive-baseline.sh`, which is skipped whenever the Hive baseline
tarball is restored (the default `refresh` mode). The Paimon HMS databases
are therefore missing on every baseline-restore run and
`test_paimon_hms_catalog` fails with `Unknown database 'hdfs_db'`. Move
the bootstrap into a `paimon_hms` module so it is reached through
`refresh-hive-modules.sh` as well; `init-hive-baseline.sh` now goes through
the same module so the HQL is not executed twice on a full init.
3. `test_es_query` moved from `external_table_p0` to `external_table_p2` but
still hardcodes `regression_test_external_table_p0_es`, while the
framework derives the database from the suite directory. Use
`context.dbName`.
4. `test_file_cache_query_limit` curls `externalEnvIp` for the backend brpc
and HTTP endpoints. That is the third-party Docker host, which runs no
backend at all in a multi-host deployment, so `file_cache_capacity` is
never found. Resolve the backend from `show backends`, and skip when more
than one backend is alive, which the case already documents as its
precondition.
5. `test_file_cache_statistics` asserts
`normal_queue_curr_size < normal_queue_max_size`. A queue's `max_size` is a
soft limit: `try_reserve_from_other_queue` lets a queue grow past its own
share while the whole cache still fits, evicting from the under-used
queues instead. Whether the assertion holds depends on how full the other
queues happen to be, which is why it is flaky. Assert the bound the backend
actually enforces, the total cache capacity.
Items 1 and 2 apply to master and branch-4.1 as well and should be forward
ported.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
|`paimon_hms`|`scripts/create_external_paimon_scripts/create_paimon_tables.hql`| Paimon tables registered in HMS (`hdfs_db`, `ali_db`, …); only selected when `enablePaimonHms=true`|
finalStringNORMAL_QUEUE_SIZE_VALIDATION_FAILED_MSG=NORMAL_QUEUE_CHECK_FAILED_PREFIX+"size validation failed (curr_size should be > 0 and < max_size)"
35
-
finalStringNORMAL_QUEUE_ELEMENTS_VALIDATION_FAILED_MSG=NORMAL_QUEUE_CHECK_FAILED_PREFIX+"elements validation failed (curr_elements should be > 0 and < max_elements)"
34
+
finalStringNORMAL_QUEUE_SIZE_VALIDATION_FAILED_MSG=NORMAL_QUEUE_CHECK_FAILED_PREFIX+"size validation failed (curr_size should be > 0 and <= total cache capacity)"
35
+
finalStringNORMAL_QUEUE_ELEMENTS_VALIDATION_FAILED_MSG=NORMAL_QUEUE_CHECK_FAILED_PREFIX+"elements validation failed (curr_elements should be > 0)"
36
36
37
37
// Constants for hit and read counts check
38
38
finalStringHIT_AND_READ_COUNTS_CHECK_FAILED_PREFIX="Hit and read counts check failed: "
0 commit comments