MariaDB 10.11.5 mtr --big results from amd64/arm64/armhf/ppc64el/riscv64/s390x
Hi! I just wanted to share the summary of mariadb-test-run with --big from all available architectures on Launchpad (https://launchpad.net/~otto/+archive/ubuntu/mariadb/+builds?build_text=3D&= build_state=3Dall): s390x Completed: Failed 13/5644 tests, 99.77% were successful. Failing test(s): innodb.row_size_error_log_warnings_3 disks.disks_notembedded type_test.type_test_double maria.maria-purge unit.conc_connection main.func_json_notembedded riscv64 Completed: Failed 6/6289 tests, 99.90% were successful. Failing test(s): disks.disks_notembedded type_test.type_test_double unit.conc_connection ppc64el Completed: Failed 9/6291 tests, 99.86% were successful. Failing test(s): disks.disks_notembedded type_test.type_test_double maria.maria-purge unit.conc_connection armhf Too many failed: Failed 19/1767 tests, 98.92% were successful. Failing test(s): federated.federatedx_versioning spider/bg.basic_sql spider/bg.spider_fixes_part spider.basic_sql spider.spider_fixes_part versioning.online versioning.partition period.versioning arm64 completed: Failed 9/6291 tests, 99.86% were successful. Failing test(s): disks.disks_notembedded type_test.type_test_double maria.maria-purge unit.conc_connection amd64 Completed: Failed 9/6291 tests, 99.86% were successful. Failing test(s): disks.disks_notembedded type_test.type_test_double maria.maria-purge unit.conc_connection As you can see, these seem to affect all architectures: - maria.maria-purge: MDEV-31349, easily repeatable failure on all architectures on Debian CI, Salsa CI and Launchpad - type_test.type_test_double: MDEV-22243 wasn't actually fixed yet, and failure visible on all architectures and platforms tested - unit.conc_connection: CONC-640/MDEV-30885 visible on all architectures tested - disk.disk{_notembedded}: MDEV-21587 Of these, Daniel Black is currently fixing the item 4 in the list (disk.disk) at https://github.com/MariaDB/server/pull/2777. For the items 1-3 help would be needed. The armhf situation does not map into a single MDEV. Do we have any Raspberry Pi enthusiasts here who would like to run the mariadb-test-run suite on their Pi and try to fix some of the failures? Overall it is nice to see nearly 6k tests pass on so many architectures! The Debian builds at https://buildd.debian.org/status/package.php?mariadb only run post-build the main suite (~1060 tests), and some architectures don't run any tests (e.g. m68k), just the build, but still nice to see the architecture support situation being very good at the moment for 10.11.5 across official and in-official Debian architectures as well :) Many thanks in particular to Daniel, Marko, Sergei, Tuukka, Faustin and Andrew for helping to hunt for fixes to the MariaDB code and tests to have wide architecture support! - Otto
participants (1)
-
Otto Kekäläinen