Hi @Jclark-ctr & @VRiley-WMF - I checked with Nicholas to get confirmation on which WMCS servers should be on single vs. dual switchports.....and the cloudvirts should be single, while the cloudosd's should be dual. Can you do an audit of what you visually see onsite, and list out any discrepancies, so we can get that worked out with WMCS and Network? Much appreciated.
Thanks,
Willy
Hey @wiki_willy , here is a list of servers that seem to be apart of the discrepancy
C8
cloudswift1001 - dual (one port is dark)
cloudvirt1027 - dual
cloudvirt1026 - dual
cloudvirt1025 - dual
cloudvirt1035 - dual
cloudvirt1034 - dual
cloudvirt1033 - dual
cloudvirt1032 - dual
cloudvirt1031 - dual
D5
cloudvirt1047 - dual (1 port is dark)
cloudswift1002 - dual (1 port is dark)
cloudvirt1042 - dual
cloudvirt1041 - dual
cloudvirt1040 - dual
cloudvirt1030 - dual
cloudvirt1029 - dual
cloudvirt1028 - dual
cloudvirt1039 - dual
cloudvirt1038 - dual
cloudvirt1037 - dual
cloudvirt1036 - dual
E4
cloudvirt1057 - dual (1 port is dark)
cloudvirt1056 - dual (1 port is dark)
cloudvirt1055 - dual (1 port is dark)
cloudvirt1054 - dual (1 port is dark)
cloudvirt1029 - dual
cloudvirt1028 - dual
cloudvirt1027 - dual
cloudvirt1027 - dual
cloudvirt1026 - dual
F4
cloudvirt1061 - dual (1 port is dark)
cloudvirt1060 - dual (1 port is dark)
cloudvirt1059 - dual (1 port is dark)
cloudvirt1058 - dual (1 port is dark)
cloudvirt1052 - dual (1 port is dark)
Awesome, thanks for working on this @VRiley-WMF. @nskaggs & @cmooney - since we have some discrepancies with the number of ports being used on these cloudvirts, should we come up with a plan/process to help us free up the second switchport on them? This will help us reclaim some switchports for new installs and server migrations. Thanks, Willy
Hey @wiki_willy indeed we made a good start to this migration early in the year (see T319184), before other priorities interrupted progress.
Having checked on the hosts I can confirm that all listed as "dual (1 port is dark)" are only using their first port. So we can remove those extra cables and free up those ports immediately. I also checked Netbox and all of them apart from cloudvirt1047 only showed a single connection there. I've fixed up Netbox for cloudvirt1047 now so we can do that one also.
Checking both the puppet config and hosts themselves it seems that 12 hosts from cloudvirt1031 - cloudvirt1042 are still using the dual setup. It should be possible for us to work with the cloud team to move them to the new setup. I will check and ensure the instances vlan is being trunked to all of them which is step one, then we can adjust the host side.
cloudvirt1025 - cloudvirt1030 are set to "offline" in Netbox, so I'm not sure what the situation with those is. They all remain unticked in T319184, so it makes sense they still have two connections, but I'm not sure what the situation is with them. @taavi do you know what the status of those hosts is?
@VRiley-WMF also heads up that cloudswift1001 and cloudswift1002 were renamed cloudlb1001 and cloudlb1002 a while back. So if there are labels with the cloudswift names they should probably be changed.
Those hosts were decom'd in T351010: decommission cloudvirt1025-cloudvirt1030.eqiad.wmnet.
This has been completed, and I pulled all the cables that were not in use. Also, renamed cloudswift1001 and cloudswift1002 to cloudlb1001 and cloudlb1002 respectively.