We should upgrade all the mediawiki layer to use debian jessie. The clusters are:
In most of these cases, it's just matter of reimaging a huge number of servers.
Number of systems to reimage:
So we need to reimage 252 servers. Videoscalers and the script/deployment servers need some preparation work before they can be reimaged, but for everything else, (the huge bulk of the machines to reimage) can be done rather mechanically using wmf-reimage right now.
Out of curiosity (and so we know what to do with MW-Vagrant) What's the timeline for this?
Keeping track of the current status in https://etherpad.wikimedia.org/p/trusty-mw-reimage
Change 307482 had a related patch set uploaded (by Volans):
Reimaging: add option to reboot after the reimage
Change 307482 abandoned by Volans:
Reimaging: add option to reboot after the reimage
Reason:
Changed approach and instead creating a Python script to be run on salt master
Change 308520 had a related patch set uploaded (by Volans):
Automation: automatically reimage host
Change 308554 had a related patch set uploaded (by Volans):
Salt: reducing permissions on the master's Job cache
Mentioned in SAL [2016-09-12T10:07:53Z] <volans> reimage mw2198, mw2199 to Jessie (again) T143536
All mw* servers in codfw with the exception of mw2152 (the video scaler) are now running jessie.
Change 310278 had a related patch set uploaded (by Volans):
Salt: fix source path typo
Change 310283 had a related patch set uploaded (by Volans):
Salt: include password module
Script wmf_auto_reimage was launched by volans on neodymium.eqiad.wmnet for hosts:
['mw2198.codfw.wmnet', 'mw2199.codfw.wmnet']
The log can be found in /var/log/wmf_auto_reimage.log.
Script wmf_auto_reimage was launched by jmm on neodymium.eqiad.wmnet for hosts:
['mw2100.codfw.wmnet']
The log can be found in /var/log/wmf_auto_reimage.log.
Completed auto-reimage of hosts:
['mw2198.codfw.wmnet', 'mw2199.codfw.wmnet']
Those hosts were successful:
['mw2199.codfw.wmnet']
To set back the conftool status to their previous values run:
confctl --quiet select 'name=mw2199.codfw.wmnet' set/pooled=yes confctl --quiet select 'name=mw2198.codfw.wmnet' set/pooled=yes
Completed auto-reimage of hosts:
['mw2100.codfw.wmnet']
Those hosts were successful:
['mw2100.codfw.wmnet']
To set back the conftool status to their previous values run:
confctl --quiet select 'name=mw2100.codfw.wmnet' set/pooled=yes
Change 310309 had a related patch set uploaded (by Volans):
Automation: improve wmf-auto-reimage
Change 310587 had a related patch set uploaded (by Volans):
Salt: use puppetmaster CNAME
Change 310680 had a related patch set uploaded (by Volans):
auto-reimage: improved output
Change 310768 had a related patch set uploaded (by Volans):
Auto-reimage: increase timeout for Icinga command
Change 311605 had a related patch set uploaded (by Volans):
Reimage: improve output in case of errors
Change 311701 had a related patch set uploaded (by Volans):
Reimage: minor improvements
What is the status of terbium? From the summary it appears to have been upgraded but the host is still a trusty.