Looking into the server spares I see:

WMF3149 and WMF3300, 2 Dell PowerEdge R310, Single Intel Xeon X3450, 8GB Memory (2) 500GB 3.5 SATA that look like they fit the bill pretty well. Unless someone objects I will acquire them for this task and start setting them up

I 've talked about this with @mark. He's against using those server spares and with good reason.

It was suggested to wait for a batch of new misc servers to arrive in eqiad and use a couple of those instead. These are tracked in a procurement ticket already so it will hopefully not add a lot of extra delay.

Ok, the cleanup of spares (via another task, they were migrated into a sheet for tracking and re-audited) has resulted in my finding a few potential systems for this. @akosiaris already stated that @mark has approved the allocation of newly aquired spares, as we cannot use out of warranty spares.

I've found some in warranty spares that are a LOT more closer to the specification than the brand new dual cpu 4 disk systems. The new systems have far more memory, cpu, and disk than requested on this task. I'm not sure if @akosiaris stating that @mark approved something counts as approval (so far the only person I know allowed to do that is @faidon ;) I could be mistaken, in either case I'll list off the potential spares, Alex can pick the one that best fits the need (rather than overprovision) and then @mark can approve. (This is not a question of trust, as I trust Alex equally as well as I trust my other team members, which is quite high!)

Once @mark approves, a few things need to happen:

• This task needs to note which system was selected
• The spares tracking sheet needs to have the row of the spare system allocated removed off the EQIAD Spare Servers tab.
• This task needs to have as sub-task created in the SRE (not hardware-requests) for the setup of the system.
  • We went with ores100X, but Alex should approve and may want to use something different.
  • That task needs to have a sub-task allocated to @Cmjohnson so he can apply the new hostname labels to the allocated systems.

I've listed all of that so if @akosiaris doesn't want to wait for me to triage this on Thursday, he can do so immediately upon getting @mark's approval.

The options for this are as follows:

Existing spares:
WMF4577 & WMF4578: Identical systems, Dell PowerEdge R420, Single Intel® Xeon® Processor E5-2450 (your spec requests a CPU of 4 cores @ 2.66, this is 8 cores @ 2.1GHz), 16GB RAM Dual 500GB SATA. The warranty on these systems is through 2017-04-30, so over a year.

Brand new dual cpu 4*4tb SATA systems:

WMF4721 & WMF4723: These are part of the order of 8 brand new misc systems. They are Dell PoweEdge R430, Dual Intel® Xeon® Processor E5- 2623 V3 (So that is 3.0 @ 4 cores, dual cpu), 32GB RAM, 4*4TB SATA. While we could use these, they seem largely overprovisioned.

However, I don't want to make the decision for you, and I had incorrectly advised we had no in warranty spares for this previously. Please review the two options above and ensure @mark approves on this task. Then you can assign to either @Cmjohnson or myself for implementation of the above, or move ahead on your own. (I've only listed all the options/steps so I don't block you on this; I have no problem if you and mark comment and assign back to me for implementation.)

In T119598#1885991, @RobH wrote:

Ok, the cleanup of spares (via another task, they were migrated into a sheet for tracking and re-audited) has resulted in my finding a few potential systems for this. @akosiaris already stated that @mark has approved the allocation of newly aquired spares, as we cannot use out of warranty spares.

Actually no I haven't said explicitly that @mark has approved the allocation of newly acquired spares. I 've clearly said it was suggested. Which is a different thing.

I've found some in warranty spares that are a LOT more closer to the specification than the brand new dual cpu 4 disk systems. The new systems have far more memory, cpu, and disk than requested on this task. I'm not sure if @akosiaris stating that @mark approved something counts as approval (so far the only person I know allowed to do that is @faidon ;) I could be mistaken, in either case I'll list off the potential spares, Alex can pick the one that best fits the need (rather than overprovision) and then @mark can approve. (This is not a question of trust, as I trust Alex equally as well as I trust my other team members, which is quite high!)

As I already said above, I did not state that @mark approved anything. Nor will I, ever.

And obviously whatever decision we take on this subject will need to be approved by @mark.

[snip]

The options for this are as follows:

Existing spares:
WMF4577 & WMF4578: Identical systems, Dell PowerEdge R420, Single Intel® Xeon® Processor E5-2450 (your spec requests a CPU of 4 cores @ 2.66, this is 8 cores @ 2.1GHz), 16GB RAM Dual 500GB SATA. The warranty on these systems is through 2017-04-30, so over a year.

These look great. I think we can move on with those.

However, I don't want to make the decision for you, and I had incorrectly advised we had no in warranty spares for this previously. Please review the two options above and ensure @mark approves on this task. Then you can assign to either @Cmjohnson or myself for implementation of the above, or move ahead on your own. (I've only listed all the options/steps so I don't block you on this; I have no problem if you and mark comment and assign back to me for implementation.)

OK, no worries. @mark can you please approve moving on with WMF457{7,8} ? Thanks!

In T119598#1885991, @RobH wrote:

• We went with ores100X, but Alex should approve and may want to use something different.

These aren't machines for running ORES itself but for the ORES Redis, so should be oresrdb1xxx?

@yuvipanda: Your reasoning seems sound to me, we'll call these oresdb1xxx.

I would suggest oresredis1xxx since they'll be used both as queue and cache machines, no 'db' type things there (we can flush them whenever)

In T119598#1932642, @yuvipanda wrote:

I would suggest oresredis1xxx since they'll be used both as queue and cache machines, no 'db' type things there (we can flush them whenever)

Sounds (mostly) good to me, I was only suggesting oresdb since it was your previous suggestion. We call other redis boxes rdb, so perhaps oresrdb1xxx?

Edit update: I misread Yuvi's initial suggestion of oresrdb. I read oresdb...

So oresrdb1xxx