Page MenuHomePhabricator

ores_classification table corrupt on enwiki labs replica labsdb1001
Closed, ResolvedPublic

Description

MariaDB [enwiki_p]> select count(*) from ores_classification where oresc_model = 21 and oresc_class = 1 and oresc_probability between 0.828 and 1;
ERROR 1030 (HY000): Got error 176 "Read page with wrong checksum" from storage engine Ari

On labsdb1001.

Other labs dbs seem fine (At the very least querying enwiki_p.oresc_classification worked from labsdb1003)

Event Timeline

Bawolff created this task.Oct 13 2017, 3:38 AM
Restricted Application added a subscriber: Aklapper. · View Herald TranscriptOct 13 2017, 3:38 AM
bd808 added a project: DBA.Oct 16 2017, 4:59 AM
bd808 added a subscriber: bd808.

Adding the DBA tag, but the best answer is probably to use the new replicas instead of labsdb1001/1003 (T142807: Migrate all users to new Wiki Replica cluster and decommission old hardware).

bd808 moved this task from Backlog to Wiki replicas on the Data-Services board.Oct 16 2017, 4:59 AM
Marostegui closed this task as Resolved.Oct 16 2017, 5:31 AM
Marostegui claimed this task.
Marostegui added a subscriber: Marostegui.

Hi,

I have fixed this and also migrated the table to InnoDB which will hopefully prevent this corruption from happening again.
But yes, the best answer is to start using the new servers :-)

mysql:root@localhost [enwiki]> select count(*) from ores_classification where oresc_model = 21 and oresc_class = 1 and oresc_probability between 0.828 and 1;
+----------+
| count(*) |
+----------+
|   147453 |
+----------+
1 row in set (3.74 sec)

But yes, the best answer is to start using the new servers :-)

/me as a cloud services newb usually just uses whatever server that $ sql enwiki_p gets me :)