Page MenuHomePhabricator

setup/install/deploy new HP restbase servers for codfw
Closed, ResolvedPublic

Description

This is the tracking task for the setup and deployment of our new codfw restless

I have receives the 6 HP servers from RT9506.
@Rob can you please provide me with the following informations:

  • server names
  • Rack location

Thanks

Event Timeline

Papaul claimed this task.
Papaul raised the priority of this task from to Medium.
Papaul updated the task description. (Show Details)
Papaul added a project: ops-codfw.
Papaul changed the edit policy from "All Users" to "Subscribers".
Papaul added subscribers: Papaul, RobH, fgiunchedi, Dzahn.
Restricted Application added a subscriber: Aklapper. · View Herald Transcript

The names of the new systems will be restbase2001-2006.

We'll want to spread these out across multiple rows/racks (similar to eqiad's deployment.) As such, I'm suggesting the following locations:

restbase2001 : b5
restbase2002 : b8
restbase2003 : c1
restbase2004 : c5
restbase2005 : d1
restbase2006 : d5

Please review each rack location for the following (I've checked these remotely, but you should do a double-check as the on-site):

  • rack has available power ports
  • rack has available power overhead (we aren't hitting 8.6kW per rack total draw)
  • rack access switch has available ports.

Once you have completed the above, please go ahead and assign mgmt dns for asset tag and hostnames (one patchset), and then later you can assign the production entries. (We can step through it as you handle the bulk of the install, or I can take it over, your call.)

@Papaul when configuring management access in bios please also make sure that "boot mode" is set to legacy bios, we've mistakenly ordered those with uefi by default (see also T112627: bios defaults on new hardware orders to track vendor defaults)
ditto for "power management" settings, although I'm not sure what we've settled on that as a default

Servers racking complete
Rack table updated
Asset tag in place
physical label in place
BIOS settings updated

restbase2001 : b5 ge-5/0/29
restbase2002 : b8 ge-8/0/8
restbase2003 : c1 ge-/1/0/13
restbase2004 : c5 ge-5/0/13
restbase2005 : d1 ge-1/0/3
restbase2006 : d5 ge-5/0/3

Note: the iLo on restbase2003 same to have a problem I can not ping it, and I can not ssh into it. I am troubleshooting the issue.

@fgiunchedi you can start setting the port on the switches. Thanks

I couldn't ping or ssh restbase2003 because the switch had a problem i had to reset the switch now everything is working.

@Papaul: I've gone ahead and set all the vlan and port descriptions, so you should be ok to continue along and put in the production dns entries.

Change 238859 had a related patch set uploaded (by Dzahn):
Add MAC address entries for restbase200[1-6]

https://gerrit.wikimedia.org/r/238859

Change 238859 merged by Dzahn:
Add MAC address entries for restbase200[1-6]

https://gerrit.wikimedia.org/r/238859

Thank you Rob I will start to work on production DNS now.

Change 239024 had a related patch set uploaded (by Filippo Giunchedi):
Add production DNS for restbase200[1-6]

https://gerrit.wikimedia.org/r/239024

Change 239024 merged by Filippo Giunchedi:
Add production DNS for restbase200[1-6]

https://gerrit.wikimedia.org/r/239024

Change 239051 had a related patch set uploaded (by Filippo Giunchedi):
install_server: add restbase200[1-6] to netboot

https://gerrit.wikimedia.org/r/239051

Change 239051 merged by Filippo Giunchedi:
install_server: add restbase200[1-6] to netboot

https://gerrit.wikimedia.org/r/239051

DNS merged, now debugging an issue where sda shows up as 1gb drive, possibly a virtual drive from ilo

~ # cat /sys/block/sda/device/model 
LUN 00 Media 0  
~ # grep sda /proc/partitions 
   8        0    1048576 sda
   8        1     947803 sda1
   8        2          1 sda2
   8        5      96358 sda5
~ #

the magic bios options to exclude the virtual media can be found under

System Configuration -> BIOS/Platform Configuration (RBSU)  -> System Options -> USB options

Embedded User Partition            [Disabled] 
Internal SD Card Slot                   [Disabled]

all new machines have been provisioned and OS installed, keys signed / puppet /etc