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
God morning,
This is a reminder that the work below will commence in approximately 15 minutes' time.
If you want to get in touch to discuss anything about this work, please feel free to use #wikimedia-analytics on IRC or #data-platforms-sre on Slack, or reply to this mail.
Kind regards, Ben
On 14/11/2023 12:13 pm, Ben Tullis wrote:
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/