[debian-mysql] Last chance to help polish MariaDB for Bookworm - mariadb 1:10.11.2-2~exp1 in experimental

Fri Mar 24 17:18:39 GMT 2023

Hi!

Following up on this:

I did add

--mysqld=--loose-innodb-flush-method=fsync \
--mysqld=--loose-skip-innodb-use-native-aio \

in https://salsa.debian.org/mariadb-team/mariadb-server/-/commit/41fb058a967c800c7578b0a5026e5dc3c43ccfad#8756c63497c8dc39f7773438edf53b220c773f67
as requested by Marko in https://jira.mariadb.org/browse/MDEV-30728
and it did fix the ppc64el test failures.

New and old logs visible at
https://buildd.debian.org/status/logs.php?pkg=mariadb&arch=ppc64el

The kernel version before and after was:
Linux 5.10.0-21-powerpc64le #1 SMP Debian 5.10.162-1 (2023-01-21)
ppc64el (ppc64le)

Above posted also to https://jira.mariadb.org/browse/MDEV-30728.

For the s390x issue below I just added
main.statistics_upgrade_not_done (MDEV-30084) for s390x skiplist.

On Sat, 18 Mar 2023 at 23:29, Otto Kekäläinen <otto at debian.org> wrote:
>
> I didn't get much response to this, but anyway looking at
> https://buildd.debian.org/status/package.php?p=mariadb&suite=experimental
> current failures are:
>
> ********************************
>
> https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=ppc64el&ver=1%3A10.11.2-2%7Eexp1&stamp=1679174850&raw=0
> Failing test(s): main.index_merge_innodb main.backup_locks
>
> main.backup_locks 'innodb'               w6 [ fail ]
>         Test ended at 2023-03-18 21:25:29
> CURRENT_TEST: main.backup_locks
> mysqltest: At line 206: query 'CREATE TABLE IF NOT EXISTS t3 (c1
> CHAR(1) BINARY,c2 SMALLINT(10),c3 NUMERIC(1,0), PRIMARY KEY(c1(1)))
> ENGINE=InnoDB' failed with wrong errno ER_UNKNOWN_STORAGE_ENGINE
> (1286): 'Unknown storage engine 'InnoDB'', instead of
> ER_CANT_UPDATE_WITH_READLOCK (1223)...
>
> main.index_merge_innodb 'innodb'         w3 [ fail ]
>         Test ended at 2023-03-18 21:24:38
> CURRENT_TEST: main.index_merge_innodb
> mysqltest: At line 86: query 'SELECT COUNT(*) FROM
> (SELECT * FROM t1 FORCE INDEX(primary,idx)
> WHERE a BETWEEN 2 AND 7 OR pk=1000000) AS t' failed: ER_NOT_KEYFILE
> (1034): Index for table 't1' is corrupt; try to repair it
>
>
> Rebuild https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=ppc64el&ver=1%3A10.11.2-2%7Eexp1&stamp=1678728871&raw=0
> failed only on main.index_merge_innodb
>
> The index_merge_innodb was reported in
> https://jira.mariadb.org/browse/MDEV-30728 and I will try using `./mtr
> --mysqld=--loose-innodb-flush-method=fsync
> --mysqld=--loose-skip-innodb-use-native-aio` in next upload. I might
> also try not using `eatmydata` as it might also affect this.
>
> ********************************
>
> https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=s390x&ver=1%3A10.11.2-2%7Eexp1&stamp=1679174792&raw=0
> Failing test(s): main.statistics_upgrade_not_done
>
> main.statistics_upgrade_not_done 'innodb' w2 [ fail ]  Found
> warnings/errors in server log file!
> -> segfault with core dump
>
> Rebuild https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=s390x&ver=1%3A10.11.2-2%7Eexp1&stamp=1678722906&raw=0
> failed on main.mysql_upgrade main.order_by_optimizer_innodb
>
> Upgrade failed on crash but optimizer test had output difference:
>
> main.order_by_optimizer_innodb 'innodb'  w2 [ fail ]
>         Test ended at 2023-03-13 15:50:29
> CURRENT_TEST: main.order_by_optimizer_innodb
> --- /<<PKGBUILDDIR>>/mysql-test/main/order_by_optimizer_innodb.result
> 2023-02-01 17:28:03.000000000 +0000
> +++ /<<PKGBUILDDIR>>/mysql-test/main/order_by_optimizer_innodb.reject
> 2023-03-13 15:50:19.568856370 +0000
> @@ -41,7 +41,7 @@
>  # The following should use range(ux_pk1_fd5), two key parts (key_len=5+8=13)
>  EXPLAIN SELECT * FROM t2 USE INDEX(ux_pk1_fd5) WHERE pk1=9 AND fd5 <
> 500 ORDER BY fd5 DESC LIMIT 10;
>  id select_type table type possible_keys key key_len ref rows Extra
> -1 SIMPLE t2 range ux_pk1_fd5 ux_pk1_fd5 13 NULL 138 Using where
> +1 SIMPLE t2 ref ux_pk1_fd5 ux_pk1_fd5 4 const 138 Using where
>  # This also must use range, not ref. key_len must be 13
>  EXPLAIN SELECT * FROM t2                       WHERE pk1=9 AND fd5 <
> 500 ORDER BY fd5 DESC LIMIT 10;
>  id select_type table type possible_keys key key_len ref rows Extra
>
> The order_by_optimizer_innodb failure is potentially related to
> https://jira.mariadb.org/browse/MDEV-30629
>
> The other issues might be related to MDEV-30728 or MDEV-30518. Help
> reading the stack traces would be useful.
>
> ********************************
>
> https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=hppa&ver=1%3A10.11.2-2%7Eexp1&stamp=1678949076&raw=0
> Fails to compile. Regression from build a couple weeks ago.
>
> Rebuild https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=hppa&ver=1%3A10.11.2-2%7Eexp1&stamp=1678729859&raw=0
> also failed to compile.
>
> ********************************
>
> https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=sparc64&ver=1%3A10.11.2-2%7Eexp1&stamp=1679195720&raw=0
> Failing test(s): main.long_unique_bugs main.implicit_commit
> main.ctype_utf8mb4_innodb main.func_json_notembedded
>
> All tests failed on timeout/segfault with core dump.
>
> Rebuild https://buildd.debian.org/status/fetch.php?pkg=mariadb&arch=sparc64&ver=1%3A10.11.2-2%7Eexp1&stamp=1678742133&raw=0
> failed on main.long_unique_bugs main.implicit_commit
> main.func_json_notembedded, same symptoms with crashes.
>
> The func_json_notembedded failure is reported in
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1030630 and
> https://jira.mariadb.org/browse/MDEV-30518 (affects multiple archs)
> but no progress on fixing it.
>
> ********************************
>
>
>
>
> On Mon, 13 Mar 2023 at 08:24, Otto Kekäläinen <otto at debian.org> wrote:
> >
> > Hi!
> >
> > I just uploaded mariadb 1:10.11.2-2~exp1 into experimental for
> > testing. If nothing surfaces, I will upload this to Debian unstable
> > next weekend and it will be the final version of MariaDB in Debian
> > Bookworm.
> >
> > If you want to help polish MariaDB in Bookworm, please help by (in
> > order of importance):
> >
> > 1. Review past commits in
> > https://salsa.debian.org/mariadb-team/mariadb-server/-/commits/debian/latest?ref_type=heads
> > and post feedback/improvement suggestions on the maintainermailing
> > list (pkg-mysql-maint at lists.alioth.debian.org) so people can see
> > how many reviews happened and what they commented on.
> >
> > 2. Review open MRs at
> > https://salsa.debian.org/mariadb-team/mariadb-server/-/merge_requests
> > about improvements that might still go into next upload, or a future
> > one.
> >
> > 3. Review genera QA system feedback at
> > https://tracker.debian.org/pkg/mariadb and post MRs to fix whatever
> > you found and want to fix yourself.
> >
> > - Otto