So in total we have 8 connections that need to run from the old cage to the new cage in Eqiad.
LVS Links
4 between the "spine" switches (we're currently using lsw1-e1 and lsw-f1 for these waiting on the QFX5120-32Cs) and the LVS servers that are being put in place. I believe we ordered the optics and cables for all of these, but we probably need to double check.
Device A | Device A Port | Device A Optic | Device B | Device B Port | Device B Optic |
---|---|---|---|---|---|
lsw1-e1 | xe-0/0/46 | 10GBase-LR | lvs1017 | ens2f0np0 - first port on pci3 | 10GBase-LR |
lsw1-e1 | xe-0/0/47 | 10GBase-LR | lvs1018 | ens2f0np0 - first port on pci3 | 10GBase-LR |
lsw1-f1 | xe-0/0/46 | 10GBase-LR | lvs1019 | ens2f0np0 - first port on pci3 | 10GBase-LR |
lsw1-f1 | xe-0/0/47 | 10GBase-LR | lvs1020 | ens2f0np0 - first port on pci3 | 10GBase-LR |
We'll also need to move all of these to the Spine switches when they are installed. So if we are ordering anything from fs.com best to make sure we have the 4x10G QSFP+ modules we need to do 4x10G (or use the QSFP->SFP+ adapters). We should probably validate and discuss on call next week.
WMCS Links
We've decided to dedicate two racks in the new cage to WMCS, and configure the switches there are more "cloudsw" devices connected directly with cloudsw1-c8 and cloudsw1-d5 in the existing cage. I've opened T301414 to make a call on what racks to assign them, but either way we'll need cables run like this:
Device A | Device A Port | Device A Optic | Device B | Device B Port | Device B Optic |
---|---|---|---|---|---|
lsw1-e4 | et-0/0/54 | 40GBaseLR | cloudsw1-d5 | et-0/0/52 | 40GBase-LR |
lsw1-e4 | et-0/0/55 | 40GBaseLR | cloudsw1-c8 | et-0/0/52 | 40GBase-LR |
lsw1-f4 | et-0/0/55 | 40GBaseLR | cloudsw1-c8 | et-0/0/53 | 40GBase-LR |
lsw1-f4 | et-0/0/54 | 40GBaseLR | cloudsw1-d5 | et-0/0/53 | 40GBase-LR |
I'm pretty sure none of the optics for these WMCS links have been purchased so we'll need to do that.
We also need to properly record the entire fiber path, with all the patch panel locations etc., and make sure it's documented correctly in Netbox (similar to open request for details of runs from new cage to CRs).