Hello,
We have to carry out some scheduled maintenance that will require a
brief period of disruption for our analytics_meta
<https://wikitech.wikimedia.org/wiki/Data_Engineering/Systems/Analytics_Meta>
MariaDB service, whilst it is moved to a new primary host
<https://phabricator.wikimedia.org/T284150>. This will affect Hive, both
Druid clusters, Superset, Hue, and DataHub.
I plan to do this work tomorrow morning, starting shortly after 11:00
UTC and I expect the change to take no more than around 20 minutes,
during which time you might find that the above services are disrupted.
Our production pipelines that write to HDFS and Hive will be while the
work is being carried out.
I will likely also put HDFS briefly into Safe Mode, which prevents write
access, whilst I reconfigure and restart Hive.
If you could plan to work your own tasks and pipelines around this
maintenance window, I would be grateful. Please do get in touch if you
have any questions, or this maintenance plan will cause you any specific
inconvenience. If you think that you will be adversely affected I can
see whether it is possible to reschedule or find another workaround for you.
Kind regards,
Ben
--
*Ben Tullis*(he/him)
Senior Site Reliability Engineer
Wikimedia Foundation <https://wikimediafoundation.org/>