A bunch of shards were retired as part of issue 753890
But these servers were left behind in server_db as primary shards, causing push-to-prod to fail.
(This would also cause our metrics dashboards to show incorrect graphs, since they will be considered to be in prod. I'd even expect shard down alerts to start firing)

Details from https://bugs.chromium.org/p/chromium/issues/detail?id=753890#c20

pprabhu@pprabhu:files$ atest server list chromeos-server42.cbf.corp.google.com
Hostname     : chromeos-server42.cbf.corp.google.com
Status       : primary
Roles        : shard
Attributes   : {}
Date Created : 2016-04-18 11:04:57
Date Modified: 2016-04-18 11:04:57
Note         : None

pprabhu@pprabhu:files$ atest server list chromeos-server43.cbf.corp.google.com                                       
Hostname     : chromeos-server43.cbf.corp.google.com
Status       : primary
Roles        : shard
Attributes   : {}
Date Created : 2016-04-18 13:13:20
Date Modified: 2016-04-18 13:13:20
Note         : None

pprabhu@pprabhu:files$ atest server list chromeos-server44.cbf.corp.google.com                                       
Hostname     : chromeos-server44.cbf.corp.google.com
Status       : primary
Roles        : shard
Attributes   : {}
Date Created : 2016-04-20 15:30:31
Date Modified: 2016-04-20 15:30:31
Note         : None

pprabhu@pprabhu:files$ atest server list chromeos-server45.cbf.corp.google.com                                       
Hostname     : chromeos-server45.cbf.corp.google.com
Status       : primary
Roles        : shard
Attributes   : {}
Date Created : 2016-04-26 11:16:28
Date Modified: 2016-04-26 11:16:28
Note         : None

These should not have been in server_db