Hey folks,
I talked with Max Spevack today about the status of the Dell server donation. To his knowledge as of yesterday nothing has arrived for him yet.
Max agreed generally that we need more boxes in the colo, and he wants information of what type of servers (features, U sizes, cost). I assume the type of serial console with access to BIOS is an important requirement, but otherwise we need standard rackmount boxes? Does anyone know any more details about our colo?
We need to get this information together, then Max can figure out if it fits within budget, and coordinate with Red Hat GIS to get it installed into the colo.
Warren Togami wtogami@redhat.com
On Tue, 15 Aug 2006, Warren Togami wrote:
Hey folks,
I talked with Max Spevack today about the status of the Dell server donation. To his knowledge as of yesterday nothing has arrived for him yet.
Max agreed generally that we need more boxes in the colo, and he wants information of what type of servers (features, U sizes, cost). I assume the type of serial console with access to BIOS is an important requirement, but otherwise we need standard rackmount boxes? Does anyone know any more details about our colo?
I don't particularly care about all the technical specs of the boxes -- I just need to know what we need in general, what we plan to use it for, and how much it's going to cost.
There shouldn't be any budget issues for a couple of thousand dollars worth of boxes, but that's only if I can give a useful accounting of what the money is buying us.
--Max
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
I am also covering the other needed costs for switching/pdus/racks/etc....
Nate and Warren are speaking now about specific HW.
As an FYI, I would like the ok to take the fedora infrastructure off line, for a day, the last week of this month. I have "acquired" (don't ask me questions on this) a full rack. I would like to move your stuff from the current half height rack gafton and I "acquired" a few years ago and move it into the full height unit. This will require about 6 hours of downtime while we move stuff. Honestly, maybe much less, but I need to be safe.
I will jump on the meeting tomorrow on IRC and also talk about it so everyone knows and approves....
Max Spevack wrote: | On Tue, 15 Aug 2006, Warren Togami wrote: | |> Hey folks, |> |> I talked with Max Spevack today about the status of the Dell server |> donation. To his knowledge as of yesterday nothing has arrived for him |> yet. |> |> Max agreed generally that we need more boxes in the colo, and he wants |> information of what type of servers (features, U sizes, cost). I |> assume the type of serial console with access to BIOS is an important |> requirement, but otherwise we need standard rackmount boxes? Does |> anyone know any more details about our colo? | | | I don't particularly care about all the technical specs of the boxes -- | I just need to know what we need in general, what we plan to use it for, | and how much it's going to cost. | | There shouldn't be any budget issues for a couple of thousand dollars | worth of boxes, but that's only if I can give a useful accounting of | what the money is buying us. | | --Max |
- -- ======================================================== = Stacy J. Brandenburg Red Hat Inc. = = Manager, Network Operations sbranden@redhat.com = = 919-754-4313 http://www.redhat.com = ========================================================
Key ID 0x1828D94D Fingerprint 03F7 43BE 1150 CCFA F57B 54DD AEDB 1C27 1828 D94D
On 8/16/06, Max Spevack mspevack@redhat.com wrote:
On Tue, 15 Aug 2006, Warren Togami wrote:
Hey folks,
I talked with Max Spevack today about the status of the Dell server donation. To his knowledge as of yesterday nothing has arrived for him yet.
Max agreed generally that we need more boxes in the colo, and he wants information of what type of servers (features, U sizes, cost). I assume the type of serial console with access to BIOS is an important requirement, but otherwise we need standard rackmount boxes? Does anyone know any more details about our colo?
I don't particularly care about all the technical specs of the boxes -- I just need to know what we need in general, what we plan to use it for, and how much it's going to cost.
There shouldn't be any budget issues for a couple of thousand dollars worth of boxes, but that's only if I can give a useful accounting of what the money is buying us.
--Max
--
As far as generals any sort of remote management would be fine by me, serial bios support, DRAC, or IP based KVM would be fine. One thing I'm not sure on, and Stace would know more, is how much space is left on our PDU's. Though in the near future we could really use a Xen box. Which, in general needs drive space and plenty of RAM.
-Mike
One more thing we need to address in the not-to-distant future is a DR plan. I'm sure we don't need anything fancy, as far as essentials we'd need our primary mirror, builders, accounting and VCS. We'd also need our primary website (wiki) to communicate with our developers and end users. At present the wiki is at Duke. I'll probably start backing it up to the colo weekly with incrementals so we can mitigate that risk if Duke goes bye bye.
I'm not sure where this is on our radar but the Phoenix colo is our primary site and if something happened to it, the Fedora Project as a whole might be in trouble. Unless there are some internal backups being done I'm not aware of. Well, except for Core. I'm not sure where the primary core cvs/look aside cache resides.
-Paranoid Mike :)
CVS is not backed up. The fedora bits released are.
As for failover, (KNOCK ON WOOD), we have not had a failure in the DC. Not to mention if we do, fedora is least of our worries honestly....
We can leverage TPA if needed and you can free up servers for it though.
Mike McGrath wrote:
One more thing we need to address in the not-to-distant future is a DR plan. I'm sure we don't need anything fancy, as far as essentials we'd need our primary mirror, builders, accounting and VCS. We'd also need our primary website (wiki) to communicate with our developers and end users. At present the wiki is at Duke. I'll probably start backing it up to the colo weekly with incrementals so we can mitigate that risk if Duke goes bye bye.
I'm not sure where this is on our radar but the Phoenix colo is our primary site and if something happened to it, the Fedora Project as a whole might be in trouble. Unless there are some internal backups being done I'm not aware of. Well, except for Core. I'm not sure where the primary core cvs/look aside cache resides.
-Paranoid Mike :)
On Wed, 2006-08-16 at 17:36 -0500, Mike McGrath wrote:
One more thing we need to address in the not-to-distant future is a DR plan. I'm sure we don't need anything fancy, as far as essentials we'd need our primary mirror, builders, accounting and VCS. We'd also need our primary website (wiki) to communicate with our developers and end users. At present the wiki is at Duke. I'll probably start backing it up to the colo weekly with incrementals so we can mitigate that risk if Duke goes bye bye.
Shouldn't our backups in general cover the disaster recovery process? and in the event of a real disaster how is the response time going to be at the coloc for drive replacement, etc?
-sv
infrastructure@lists.fedoraproject.org
-
Max Spevack -
Mike McGrath -
seth vidal -
Stacy J. Brandenburg -
Warren Togami