1088569 - (temporary) VMs for upcoming Elasticsearch test & migration

Status: RESOLVED FIXED

(Infrastructure & Operations :: Virtualization, task)

Description

[Daniel Maher [:phrawzty]]

Hello,

We will require a three VMs as part of our Q4 goal to migrate Socorro to Elasticsearch 1.x.  These VMs will be temporary - as soon as the test and migration period is complete, they will no longer be required.  The CPU requirement is pretty low (the real metal machines have way more processors than they need), but we do have relatively steep RAM and HDD requirements - I don't know what's reasonable, so please tell me if this is ridiculous.

---

Desired hostname(s):
socorro-es[4-6].stage.phx1.mozilla.com

Number of cores:
2

Amount of RAM:
12 GB

Desired OS:
RHEL x86_64

Amount of disk:
1 TB

Comment 1

[Daniel Maher [:phrawzty]]

In the event that this is not feasible (storage req, for example), then I'd gladly accept suggestions as to how we could acquire equivalent bare metal for this purpose, heh. ;)

Comment 2

[Chris Knowles [:cknowles]]

As dmaher implies, we've been talking about this.

2CPU - no problem
12G RAM - a little large, but for temporary VMs and with a history of needing that much ram, totally possible.
1TB each - Yeahhh... that's a little much.  However, as long as these *really* are temporary, we can swing it.  If you can give us a drop dead date that we'd have to support the extra 3TB in the datastores, then we can move forward.

Only normal questions we have:
hostname: I think you meant to stick a webapp in there, socorro-es{4-6}.stage.webapp.phx1.mozilla.com - correct me if I'm wrong
Disk - do you want that giant bolus of disk as part of / or a different partition?

So, give me a date where I can come and collect my VMs or some teeth, and those other questions, and we'll try to get these up for you.  (likely early next week at this rate - unless this is drop dead NOW priority)

Comment 3

[Chris Knowles [:cknowles]]

Haven't heard from you as to date of VM removal and these other questions:
hostname: I think you meant to stick a webapp in there, socorro-es{4-6}.stage.webapp.phx1.mozilla.com - correct me if I'm wrong
Disk - do you want that giant bolus of disk as part of / or a different partition?

Let me know so we can move forward on these.

Comment 4

[Daniel Maher [:phrawzty]]

Sorry - I was out of pocket for the past couple of days.  To answer your questions:

* Yes, the "webapp" string should be in the hostname: socorro-es{4-6}.stage.webapp.phx1.mozilla.com .
* Regarding the disk, a giant bolus is fine.
* Since this is related to a Q4 goal, and since having a hard cut-off will help force us to finish (heh), let's put the expiry at the end of this quarter (with padding): so 9 January 2015.
* Whenever you get around to this is fine (within reason, heh).  Thanks !

Comment 5

[Chris Knowles [:cknowles]]

Alright, those are up.  3 VMs with insanely large / volumes.  

Initial puppet and Nagios have been applied, ready for your customization.

I've also dropped in a note on my calendar to come calling around January 2015.

Let me know if there's any problems.

Comment 6

[Adrian J Fernandez [:Aj]]

I removed the parent "esx-cluster1.private.corp.phx1.mozilla.com" as these hosts were added to the non corp-phx1 nagios configuration.

If these VMs are indeed in our "corp" (ie used to be HCI set-up), they would need to be moved to corp-phx1 nagios.

Comment 7

[Daniel Maher [:phrawzty]]

Feel free to decom these machines at your leisure.

Comment 8

[Chris Knowles [:cknowles]]

Alright, those are out of nagios and off, and we're waiting the standard week for cold feet, and then I'll fully decomm them.

Comment 9

[Chris Knowles [:cknowles]]

And now they're out of puppet/rhn and inventory.  and removed from disk.