10 new systems were ordered on parent task T154698. Those 10 systems will replace the following systems:
>>! In T152698#2903671, @Joe wrote:
> My suggestion would be that these 10 new systems should replace mw2075 - mw2089 functionally, and specifically:
>
> - **3 servers** to replace the 5 API appservers mw2075-79 so in **row A**
> - **4 servers** to replace 6 jobrunners mw2080-2085 so in **row B**
> - **3 servers** to replace 4 imagescalers mw2086-2090 so in **row B**
>
Please note that while @joe listed up to mw2290, it did not include it, as it isn't an imagescaler. (He also lists 4 image scalers to be decomed, but range includes 5 systems, 4 imagei and 1 general.)
After further discussion on that task, @joe pointed out that the old systems can come offline in advance of the new ones. I'll (@robh) will be pulling them for decommission today/tomorrow. Then @papaul will take over for the on-site steps of wiping the disks and unracking the systems.
Please note when these systems are unracked, @papaul may want to leave all cables in place for the new systems that will eventually go in those spots.
Servers for decommission: mw2075-2079 (api), mw2080-2085 (job), mw2086-2090 (image)
Steps for each system in decommissioning:
[x] - disable all service level checks in icinga for hosts (done for all hosts mw2075-20989 set in maint/downtime)
[x] - depool from pybal (sudo -i confctl select name=,<fqdn hostname> set/pooled=no (done for all hosts)
[] - disable puppet on hosts
[] - remove from puppet, includes: conftool-data, install_server, hiera - https://gerrit.wikimedia.org/r/#/c/330621/
[] - system shutdown
[] - pull production dns entires
[] - disable network port
[] - puppet node clean and deactive
[] - salt key revoked
[] - hand off system to @papaul for disk wipe.
[] - disk wiped
//Please note that the mgmt dns entires and the network port description are not removed until AFTER system is unracked.//
[] - remove mgmt dns entires
[] - remove description from switch port config