Hi,
all analytics slaves are working again.
On Tue, Apr 22, 2014 at 12:33:09AM +0200, Christian Aistleitner wrote:
The s7 slave is already being unreachable at the time of writing.
s7-analytics-slave.eqiad.wmnet has been updated to point to a different machine, and is usable again.
Thanks to springle for the quick fix!
But Ops also said that anything hammering db1007 too hard will be hastily killed. So please do not run unneeded slow reports on the machine.
Ops are working on getting db68 up again and then make s7-analytics-slave.eqiad.wmnet point to db68 again.
I guess due to the pmpta servers being taken offline. And I expect the remaining ones in the above list to go down soon as well.
That warning turned out to be overcautious. The db68 issue turned out to be an unrelated issue that just happened to occur on the day Ops turned off some pmtpa machines.
Ops told us that the analytics slaves are not going away yet because they are on a different floor of the data center.
Have fun, Christian