#4948 January 2016 phx2 on-site vist / work
Closed: Fixed None Opened 3 years ago by kevin.

We will be having an on site visit to the phx2 datacenter in january of 2016

This visit is tenatively planned for the week of January 11th, 2016

This ticket is to capture what work is planned.

  1. Move / consolidate systems in rack 47 in particular to allow cloud expansion into it (from rack 46).

  2. Rack and configure HP moonshot system in rack 16.

  3. Assist RHIT with replacement of top of rack switches in all Fedora racks.

  4. Move QA/Secondary Arch and other machines in the 'qa' networks to seperate networks.

  5. Possibly firmware updates on various systems.


We are currently trying to get a site visit together to PHX2 to clean up
various things. Our current planned time to be there is from January
11th to possibly the 16th.

The following tentative agenda is with the assumption that we do not
have an extra rack to put systems into. If that changes we can move this
around separately.

  1. Clean up QA racks. '''Reason''': current racks are close to power capacity. '''Tasks''':
    A. Move PPC systems over to PPC rack 17
    B. Move arm systems to arm rack rack 16
    C. Move some QA systems in rack 47 to rack 51,52

  2. Clean up cloud racks. '''Reason''': current rack is over power capacity and we need to split the load over 2 racks. Requirements: we will need to clean out rack 47 to make room for cloud items. We will need either wiring from rack 46 to 47 or an additional switch.

  3. Clean up wiring in racks 48->52. '''Reason''': wiring in these racks has been done in getitdone mode versus planned amounts. getting the wiring cleaned up in each rack with measured and run amounts will allow for the switch assignments and kvm and serial assignments to be better documented and fit.

'''Subtasks''':
1. Make KVM wiring only to systems that require logins from external groups.
A. kernel systems
B. sigul systems?
C. gnome systems
D. specific qa systems?
2. Make serial wiring only to systems that require serial access.
A. PPC/arm systems
B. QA/beaker systems
3. Other items needed to be done?

'''Alternative plan''': If we are able to get an empty rack we will do a move
and slide where items in one rack (starting with 52 and 48) are moved to
the empty rack. We then slide 51 to 52 with cleaned up wiring, 50 to 51,
etc etc until rack 47 is clear and the cloud items in 46 can be moved to it.

Is 4 from the original ticket filed still likely to happen? If so, will it be during the site visit and is there more planning/prep that needs to happen from the qa/secarch folks?

We can try and do it then (would be a good time as we will have networking folks on hand to make changes for us), or we could decide to defer it.

Do you have a preference?

I think as far as prep goes we should probibly come up with a document that gives what exact machines will move to/from what ips/vlans. That would allow us to move through them one at a time and make sure they are moved to the place we want them to be. Perhaps something in gobby we can all edit?

item 3 from the orig ticket discription (replace top of rack switches) has been deferred. Item 5 is being done out of band and doesn't need hands on work.

So, revised list:

  1. Move equalogics02 from rack 51/u25 to top of rack 47.

  2. Move some not yet determined power7/8 box to top of rack 47

  3. Have new switch at top of rack 47 added to connect with cloud rack (46) for above 2 items.

  4. rack and get working HP moonshot in rack 16

  5. move arm01/02/04 from rack 47 to rack 16

  6. connect new kvm in rack50 to machines we want to connect to it.

  7. connect serial to machines that need it.

  8. prep rack 59 for more machines down the road.

  9. Move PPC hosts from rack 47 and 48 to rack 17 (meeds outage of ppc koji)

  • ppc-comm04 to rack 17 U14-10
  • ppc-comm03 to rack 17 U09-04
  1. add memory to cloud machines (needs cloud outage)

  2. move TBD qa network machines vlan/network. (need to plan out exactly what machines go where this week).

  3. virthost05/06/07 could be removed or moved as they go out of warentee later this year (05 still has db-fas01 on it, needs moving/outage).

  4. re-cable/clean up wiring on all racks in general. (possible short outages).

  5. Confirm docs are up to date with racks info, pdu info and serial info, etc.

Will try and iron out the items above to be determined, then look at scheduling per day work before next week.

ok, today's revised version...
1. Move equalogics02 from rack 51/u25 to top of rack 59.·

  1. move arm03 to rack 59

  2. Move ppc-comm07/08 boxes (once things are moved off them) to top of rack 59

  3. rack and get working HP moonshot in rack 16

  4. move arm01/02/04 from rack 47 to rack 16

  5. connect new kvm in rack50 to machines we want to connect to it.

  6. connect serial to machines that need it.

  7. Pull old ppc power7 stuff

  • ppc-comm04
  • ppc-comm03
  • buildppc
  1. add memory to cloud machines. Outage wed morning at 10am.

  2. virthost05/06/07/09/10 Are going to be used for storage testing.
    Move them all to one place, like bottom of rack59

  3. re-cable/clean up wiring on all racks in general. (possible short outages).·

  4. Confirm docs are up to date with racks info, pdu info and serial info, etc.

I'd like to request that beaker-client01.qa and beaker-client02.qa be hooked up to a serial console. Beaker wants to add serial logs to jobs post-install and I'd like to allow that one the bare metal beaker clients if possible

This visit was completed.

Login to comment on this ticket.

Metadata