Closed Bug 761596 Opened 9 years ago Closed 9 years ago

Please clone sea-win32-0{1,3,4} from sea-win32-02 physical machines

Categories

(Infrastructure & Operations :: DCOps, task)

x86_64
Windows 7
task
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: Callek, Assigned: van)

Details

(Whiteboard: scl3)

So in a chat with MaRu on IRC a bit over a week ago, it was said that I might be able to get some of my iX machines recloned.

sea-win32-02 as the base-machine

sea-win32-0{1,3,4} as the "clone onto me" machines.

These are currently racked in scl3 with both normal networking and an IPMI hookup.

I am told by :dustin, that scl3 may not have the necessary tools/infra to do this clone there, but if you are willing to transfer stuff to scl1/2 to clone then back to scl3 that is fine by me (even if a few day turnaround, but not ok if its week[s] turnaround from when it is pulled from scl3)

sea-win32-02 is *currently* in production, so I'd need at least a 24 hour heads up to pull it and clear out some large disk usage on one of its drives (to speed up the cloning, aiui).

If this is done it needs racking in the same ports/racks/etc. in scl3, to accomodate current inventory/DNS/etc listings, and will need hostname/etc corrected properly on the cloned machines.

WONTFIX if this is impossible to perform, but I ask because I had spent >35 hours bringing up sea-win32-02 and I had lost the docs for what I did/investigated. (due to confusion on duration of a storage duration for a temp-storage mozilla-owned-site) And I hope to not spend that long all over again.
scl3 does not have the necessarily tools to do this (nor will it ever since we're planning on moving to a new method of installation before we move any windows hardware into scl3).  

The only way to accomplish what you ask would be to physically move the machines to scl1 or mtv1 to image them there, and then have someone sanitize them and then move them back to scl3.  This is not impossible to do, but due to the current relops workload and all of the new platforms coming online, is not likely to get done in the next month or two.
(In reply to Amy Rich [:arich] [:arr] from comment #1)
> is not likely to get done in the next month or two.

From IRC convo, it seems like we *can* streamline this delay if RelEng (CC'ed coop/joduinn) are willing to sign off on me having access to do the imaging myself (even if temporary).

And if DCOps updates DHCP/DNS/Inventory themselves this removes the need for relops to be involved, if not then its still a long wait, just not *as* long.

So Action Items:
* RelEng, confirm whether or not they are ok with me having said access
* DCOps, identify what they can/won't do, and timeframes they can commit to.
(In reply to Justin Wood (:Callek) from comment #2)
> From IRC convo, it seems like we *can* streamline this delay if RelEng
> (CC'ed coop/joduinn) are willing to sign off on me having access to do the
> imaging myself (even if temporary).

Actually CC'ing them now.
What does that access entail?
What exists in scl1/mtv1 that is missing in scl3? Would it be easier to move the cloning machine to scl3 temporarily?
(In reply to Derek Moore from comment #5)
> What exists in scl1/mtv1 that is missing in scl3? Would it be easier to move
> the cloning machine to scl3 temporarily?

It would not be easier to move one of the cloning machines (mac minis), no. We use them pretty much every day and they would not function in scl3, anyway.
We'll coordinate getting these servers moved to scl1. What can we do in the future to make sure this is something that won't happen again? Moving servers back and forth between physical locations is a significant drain on resources.
At dustin's suggestion, we might try doing this process semi-manually with Norton Ghost (or some such) inside SCL3.

Sal, let me know when you have time to talk it out.
(In reply to Derek Moore from comment #8)
> At dustin's suggestion, we might try doing this process semi-manually with
> Norton Ghost (or some such) inside SCL3.
> 
> Sal, let me know when you have time to talk it out.

I have time, whenever it works best for you guys.
(In reply to Derek Moore from comment #8)
> At dustin's suggestion, we might try doing this process semi-manually with
> Norton Ghost (or some such) inside SCL3.
> 
> Sal, let me know when you have time to talk it out.

I highly suggest http://clonezilla.org/clonezilla-live.php , it can boot from ISO or USB and supports writing to USB, Network shares (if you give the booted Linux an IP), and it handles NTFS nicely. I've used it to clone, recover, and build boxes numerous times over the last 5 years with great success.
Whiteboard: scl3
:callek,

we've picked up an external drive and have tested clonezilla (thanks solarce). i can back up your image monday if you have time to clean up sea-win32-02 over the weekend.

thanks,
van
Van,

Sure we will cleanup by sunday midnight PDT. Let me know when -2 is back up, and when -1,3,4 are also up. Thank You!
(In reply to Van Le from comment #11)
> :callek,
> 
> we've picked up an external drive and have tested clonezilla (thanks
> solarce). i can back up your image monday if you have time to clean up
> sea-win32-02 over the weekend.
> 
> thanks,
> van

awesome! glad it's working
justin,

the host is still up. am i free to shut it down?

[vle@admin1b.private.scl3 ~]$ fping sea-win32-02.community.scl3.mozilla.com
sea-win32-02.community.scl3.mozilla.com is alive

thanks,
van
i have reimaged sea-win32-[01,03,04] with sea-win32-02's image, reconfigured their hostname and network settings. please let me know if there are any issues.

[vle@admin1b.private.scl3 ~]$ fping sea-win32-{01,03,04}.community.scl3.mozilla.com
sea-win32-01.community.scl3.mozilla.com is alive
sea-win32-03.community.scl3.mozilla.com is alive
sea-win32-04.community.scl3.mozilla.com is alive

thanks,
van
Assignee: server-ops → vle
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
(In reply to Van Le from comment #15)
> i have reimaged sea-win32-[01,03,04] with sea-win32-02's image, reconfigured
> their hostname and network settings. please let me know if there are any
> issues.
> 
> [vle@admin1b.private.scl3 ~]$ fping
> sea-win32-{01,03,04}.community.scl3.mozilla.com
> sea-win32-01.community.scl3.mozilla.com is alive
> sea-win32-03.community.scl3.mozilla.com is alive
> sea-win32-04.community.scl3.mozilla.com is alive

Woot!  Thanks Van!
fyi, i noticed that these hosts are using google's dns so if you have any services that arent resolving, it might be due to the external dns.
Product: mozilla.org → Infrastructure & Operations
You need to log in before you can comment on or make changes to this bug.