Looking at T297083: [ceph] Getting rack level HA and subsequent servers moving tasks reminded me of this.
From Netbox, then manually curating the list to remove any hosts in the private or public vlans.
That leaves us with the cloud-hosts and cloud-storage WMCS servers still present in eqiad row B. Those were not moved on day 1 (when creating C8/D5) as they were too many at the time.
Moving those hosts to the dedicated WMCS racks would allow us to stop stretching the cloud-hosts and cloud-storage vlans between those racks. And remove the link connecting row B to C8 (and removing this now "snowflake").
| Name | Rack | Purchase date |
|---|---|---|
| cloudvirt1017 | B7 | 2017-05-05 |
| cloudvirt1020 | B7 | 2017-07-21 |
| cloudvirt1019 | B4 | 2017-07-21 |
| cloudvirt1022 | B7 | 2017-12-19 |
| cloudvirt1021 | B4 | 2017-12-19 |
| cloudvirt1024 | B2 | 2018-06-06 |
| cloudvirt1023 | B7 | 2018-06-06 |
| cloudcephmon1001 | B7 | 2019-05-15 |
| cloudvirt-wdqs1001 | B3 | 2019-10-16 |
| cloudvirt-wdqs1002 | B5 | 2019-10-16 |
| cloudvirt-wdqs1003 | B6 | 2019-10-16 |
All the ones purchased in 2017 and 2018 are due for a refresh, so no need to move them as their replacement will arrive in the proper racks. Is there a task tracking those decom/replacements?
That leave us with the 4 bolded hosts, so I'm wondering if cloudcephmon1001 could be moved as part of T297083: [ceph] Getting rack level HA and the cloudvirt-wdqs100x hosts when possible.