Page MenuHomePhabricator

setup/install restbase-dev100[123]
Closed, ResolvedPublic

Description

This task will track the setup of the three new restbase testing hosts, restbase-dev100[123].

This task should be reviewed and any needed corrections/adjustments from serivces. (Ensuring we partition and install as needed.)

These are out of warranty R720xd systems that will be retrofitted with LFF to SFF adapters and new SSDs in 4 of their 12 disk bays. Setup then should continue to partition and format these similar to how restbase production hosts are setup. All the extra disks (non-ssd) don't need to be setup.

The systems currently reside in three different racks and rows, which is ideal. Services preferred if none of them share a rack.

Steps for each system are identical:

restbase-dev1001 (WMF4086):

  • - update racktables and physical label of host
  • - remove first 4 HDD bays and replace with Intel S3610 800GB SSDs. SSDs were ordered on T150968, and adapters ordered on T150480
  • - setup network port, internal vlan
  • - check all bios/drac settings
  • - update and test dns entries (mgmt and production) https://gerrit.wikimedia.org/r/#/c/327051/
  • - install_module update (dhcp and partman) https://gerrit.wikimedia.org/r/#/c/327118/
  • - 4th SSD not showing fixed via sub-task T153139
  • - set all ssds in individual raid0 hw arrays, so their sw raid setup is identical to production and other restbase-test hosts.
  • - install os (jessie)
  • - accept/sign puppet/salt
  • - hand off to services for implementation

restbase-dev1002 (WMF4091):

  • - update racktables and physical label of host
  • - remove first 4 HDD bays and replace with Intel S3610 800GB SSDs. SSDs were ordered on T150968, and adapters ordered on T150480
  • - setup network port, internal vlan
  • - check all bios/drac settings
  • - update and test dns entries (mgmt and production) https://gerrit.wikimedia.org/r/#/c/327051/
  • - install_module update (dhcp and partman) https://gerrit.wikimedia.org/r/#/c/327118/
  • - set all ssds in individual raid0 hw arrays, so their sw raid setup is identical to production and other restbase-test hosts.
  • - install os (jessie)
  • - accept/sign puppet/salt
  • - hand off to services for implementation

restbase-dev1003 (WMF4093):

  • - update racktables and physical label of host
  • - remove first 4 HDD bays and replace with Intel S3610 800GB SSDs. SSDs were ordered on T150968, and adapters ordered on T150480
  • - setup network port, internal vlan
  • - check all bios/drac settings
  • - update and test dns entries (mgmt and production) https://gerrit.wikimedia.org/r/#/c/327051/
  • - install_module update (dhcp and partman) https://gerrit.wikimedia.org/r/#/c/327118/
  • - set all ssds in individual raid0 hw arrays, so their sw raid setup is identical to production and other restbase-test hosts.
  • - install os (jessie)
  • - accept/sign puppet/salt
  • - hand off to services for implementation

Event Timeline

RobH added parent tasks: Unknown Object (Task), Unknown Object (Task).

The SSDs in restbase are typically placed into a large raid0. Please copy the partitioning of restbase-test200* for these.

I'm stealing the remainder of this task from Chris. I'll update and get these installed later today.

This comment was removed by RobH.

It seems that restbase-test1001 doesn't see all of its SSDs, as it only sees 3 of the 4. The other two hosts see all four fine.

I'm creating a sub-task for Chris to investigate.

It appears the LFF to SFF adapter was the culprit for the disk not presenting itself. Replaced the adapter and all disk show up. During the process PSU2 blew and the servers is now down to 1 PSU (restbase-test1001). Also, worth noting restbase-test1003 is down to 1 PSU. I will check to see if we have any decom R720's to use for spares.

RobH updated the task description. (Show Details)

All systems are online and ready for use.

The power issue is being tracked via sub-task, resolving this setup task. Services can now take these for use.

fgiunchedi claimed this task.
fgiunchedi subscribed.

The restbase-test1* hostnames are misleading since restbase-test2* is already there but other purposes, will rename to restbase-dev

Change 330584 had a related patch set uploaded (by Filippo Giunchedi):
Rename restbase-test1* to restbase-dev1*

https://gerrit.wikimedia.org/r/330584

Change 330588 had a related patch set uploaded (by Filippo Giunchedi):
install_server: rename restbase-test1* to restbase-dev1*

https://gerrit.wikimedia.org/r/330588

Change 330584 merged by Filippo Giunchedi:
Rename restbase-test1* to restbase-dev1*

https://gerrit.wikimedia.org/r/330584

Change 330588 merged by Filippo Giunchedi:
install_server: rename restbase-test1* to restbase-dev1*

https://gerrit.wikimedia.org/r/330588

Change 330592 had a related patch set uploaded (by Filippo Giunchedi):
install_server: rename restbase-test1* to restbase-dev1*

https://gerrit.wikimedia.org/r/330592

Change 330592 merged by Filippo Giunchedi:
install_server: rename restbase-test1* to restbase-dev1*

https://gerrit.wikimedia.org/r/330592

fgiunchedi renamed this task from setup/install restbase-test100[123] to setup/install restbase-dev100[123].Jan 4 2017, 11:36 PM
fgiunchedi updated the task description. (Show Details)

It is indeed, resolving!