3/2017 Network Recable

On 3/25/2017 to 3/26/2017 nrbrase, amf42, jamesdro recabled the entire server room. Woah! Check below for the sick before and after pictures. Ha Ha Ha, just kidding we didn't take pictures.

(pictures coming soon)

• Simple ping sweep for all VM hosts
  • for i in `seq 1 17`; do ping cdr-vm$i.cse.buffalo.edu -c 1 -t 1 > /dev/null && echo "vm$i up" || echo "vm$i down"; done;
• Shutdown cdr-vcenter1 last; need to turn off the MD3000's first
• DON'T FORGET TO SAVE THE RUNNING CONFIG ON THE SWITCH GOSH DARN IT
  • System → File Management → Copy Files
• KVM Console wasn't designed for a rack. It'll be located between 21 and 22 on Rack #1. We jerry-rigged it to some rails, so it can atleast self support itself. Removing it will require removing pretty much every server above it, as it can't slide out. Sorry. It's better than what it was before, though.
  • Thank Aaron for doing it
• KVM console was put on a 3U half shelf. Pretty sick, but it's free floating as well. The spot above it is dead space, currently. It occupies units #21-23 on Rack #1
• One of the R710's (vm7 or vm11) was put on a half-shelf, as they don't have rails. The other R710 was put ontop of the shelf.
• The 2nd Rack's UPS cord is too short. We couldn't find an extension cable for it either, woop :(. It'll stay on unit #5 and 6.
• We couldn't find rails to fit on vm6's R2950. We had rails from a 2850, however one of the three pins on it does not fit. We're going to attempt to grind off the first pin, so it can be held up by two of the rails
• New color mappings decided
  • GREEN/YELLOW = Management Network
  • RED = Core vDS
    • Trunk between blue / red team nets
  • PURPLE = Team vDS
    • Trunk between generic team network, and team #1 → #15 nets
  • ORANGE = IO
  • BLUE = UNUSED
  • BLACK = Used by the AMD (PLAYGROUND) hosts. Unholy trunk between io / blue / red / vds generic net
• As most servers have 4 port mappings, we have decided upon the following order
  • Management Network
  • Core vDS
  • Team vDS
  • IO
• Servers which have 6/8 port mappings will be unused, until decided upon
• All storage servers have a LAGG between 3 ports
• The following classification of servers will use the switch's management network connection
  • Storage servers (cdr-iscsi{1-3})
  • MD3000 Management Uplinks
  • [Possible - TBD] PLAYGROUND hosts
• TODO: We will have to change the mappings on switch #2 to support the above changed port mappings. TBD tomorrow

• New color mappings
  • GREEN = Management Network, connected directly to a ZBT
  • BLUE = Management Network, connected to our switch (cdr-switch)
  • RED = Core vDS
    • Trunk between blue / red team nets
  • PURPLE = Team vDS
    • Trunk between generic team network, and team #1 → #15 nets
  • ORANGE = IO
  • BLACK = Used by the AMD (PLAYGROUND) hosts. Unholy trunk between io / blue / red / vds generic net
• Found out the UPS on Rack #2 is dead. Removed it.
• Power issues fixed. Only one power strip is connected to UPS. Storage servers should all be on this UPS-backed power strip
• Rack #1 unit #10 IS SCREWED IN. The right-back side of the rails is broken, requiring a screw.
• BE VERY CAREFUL: cdr-iscsi1 is free floating - no rails :(
• Power cable color codes
  • Red = small
  • Blue = medium
  • Yellow = large
• Power uplinks (top of racks)
  • Order: Blue / Blue / Yellow
  • Blue = ??
  • Yellow = ??