Tonight's maintenance on db9 is completed, during which it was read-only
for 7 minutes. I'm going to perform a second round of maintenance tomorrow
at the same time (Thursday 18:00PST) which will provide the long-term fix
to db9's woes. Availability of the same set of services will be
interrupted for a similar length of time.
-Asher
On Mon, Dec 19, 2011 at 5:17 PM, Asher Feldman <afeldman(a)wikimedia.org>wrote;wrote:
Hi,
We need around 20 minutes of downtime to all services that write to db9
for replication maintenance. Outside of services that support the ops
team, this primarily means bugzilla, etherpad, and civicrm. It will remain
available for read queries, however, so read usage of services such as the
tech blog should continue along fine. I'm planning to start this on Weds
at 18:00 PST and will send follow-up mail at the start and completion of
maintenance.
-Asher