Just a reminder, we will be taking the Analytics Hadoop Cluster offline
tomorrow morning EST. I’ll email again tomorrow right before we do, and
also once it is back up.
-Andrew & Luca
On Tue, Feb 14, 2017 at 2:31 PM, Andrew Otto <otto(a)wikimedia.org> wrote:
We are planning an upgrade of the Hadoop cluster on February 28th. We
need to take the cluster down for this upgrade. The actual upgrade
shouldn’t take more than 2 hours, but we’re going to reserve the whole work
day of February 28th to do this, just in case something goes wrong.
Hadoop, Hive and Spark will all be unavailable during this time. (Kafka,
Druid, pivot, and stat* boxes will continue to be accessible.)
You can keep track of our progress in Phabricator: https://
If you have a hard need to use Hadoop on this day, let us know soon. We
can reschedule this upgrade for another date if we need to.
-Andrew & Luca, your friendly A-Team Ops Engineers :)