[SAC] OSGeo4 VM revamp Plan

Alex_M · September 19, 2014, 9:27pm

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

apache · September 19, 2014, 10:39pm

Replies inline

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

We can do that; perhaps we'll make a child ticket with Spott regarding this.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

It may take a while to gnt-instance modify -t drbd though.

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

This should be fine on a temporary basis while you rebuild your ganeti node in your prefered configuration.

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

We're standardizing on CentOS; your current VMs run Gentoo and it's likely your team would fare better after the switch. I doubt we'll be ready to manage ganeti nodes with Chef by then, so it's likely to remain cfengine managed.

Step 5.
Benchmark to make sure the disks are performing ok.

Bonnie++ is the standard OSS tool for this. I find Berku's Performance Whack-A-Mole tutorial[1] covers the topic well. Though it is pgsql oriented, the discussion about the lower layers remains relevant.

Is the benchmarking step something you were wanting OSL to perform? I suspect your team knows the hardware and applications better than us.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

I'll get you that plan next week.

Justin

[1]: http://media.postgresql.org/pgcon2009/tutorials/performance.mov

Alex_Mandel1 · September 19, 2014, 11:06pm

On 09/19/2014 03:39 PM, Justin Dugger via RT wrote:

Replies inline

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

We can do that; perhaps we'll make a child ticket with Spott regarding this.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

It may take a while to gnt-instance modify -t drbd though.

gnt-instance modify -t drbd --no-wait-for-sync
So you can get it started and boot it back up.

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

This should be fine on a temporary basis while you rebuild your ganeti node in your prefered configuration.

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

We're standardizing on CentOS; your current VMs run Gentoo and it's likely your team would fare better after the switch. I doubt we'll be ready to manage ganeti nodes with Chef by then, so it's likely to remain cfengine managed.

Step 5.
Benchmark to make sure the disks are performing ok.

Bonnie++ is the standard OSS tool for this. I find Berku's Performance Whack-A-Mole tutorial[1] covers the topic well. Though it is pgsql oriented, the discussion about the lower layers remains relevant.

Is the benchmarking step something you were wanting OSL to perform? I suspect your team knows the hardware and applications better than us.

Either, just want to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

I'll get you that plan next week.

Justin

[1]: http://media.postgresql.org/pgcon2009/tutorials/performance.mov

Thanks,
--
Alex Mandel
http://wildintellect.com

apache · September 19, 2014, 11:11pm

On 09/19/2014 03:39 PM, Justin Dugger via RT wrote:

Replies inline

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

We can do that; perhaps we'll make a child ticket with Spott regarding this.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

It may take a while to gnt-instance modify -t drbd though.

gnt-instance modify -t drbd --no-wait-for-sync
So you can get it started and boot it back up.

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

This should be fine on a temporary basis while you rebuild your ganeti node in your prefered configuration.

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

We're standardizing on CentOS; your current VMs run Gentoo and it's likely your team would fare better after the switch. I doubt we'll be ready to manage ganeti nodes with Chef by then, so it's likely to remain cfengine managed.

Step 5.
Benchmark to make sure the disks are performing ok.

Bonnie++ is the standard OSS tool for this. I find Berku's Performance Whack-A-Mole tutorial[1] covers the topic well. Though it is pgsql oriented, the discussion about the lower layers remains relevant.

Is the benchmarking step something you were wanting OSL to perform? I suspect your team knows the hardware and applications better than us.

Either, just want to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

I'll get you that plan next week.

Justin

[1]: http://media.postgresql.org/pgcon2009/tutorials/performance.mov

Thanks,
--
Alex Mandel
http://wildintellect.com

neteler · October 20, 2014, 12:20pm

Hi Alex, Martin,

are there any news concerning the VMs?

thanks & sorry to bother,
Markus

neteler · November 1, 2014, 9:33am

Justin,

On Sat, Sep 20, 2014 at 12:39 AM, Justin Dugger via RT
<support@osuosl.org> wrote:

Replies inline

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

[...]

Please get back to us with a schedule for who and when we might do each
of these steps.

I'll get you that plan next week.

Do you have any news for us?

thanks,
Markus

apache · November 1, 2014, 9:33am

Justin,

On Sat, Sep 20, 2014 at 12:39 AM, Justin Dugger via RT
<support@osuosl.org> wrote:

Replies inline

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

[...]

Please get back to us with a schedule for who and when we might do each
of these steps.

I'll get you that plan next week.

Do you have any news for us?

thanks,
Markus

apache · November 25, 2014, 7:58pm

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

--
Justin Dugger
Senior System Administrator
OSU Open Source Lab

Alex_M · November 25, 2014, 11:45pm

If you still have the resources to loan us during the shuffle the
existing plan is still fine. We have not worked on future procurement at
all since our Foss4g conversation.

Thanks,
Alex

On 11/25/2014 11:58 AM, Justin Dugger via RT wrote:

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

apache · November 25, 2014, 11:46pm

If you still have the resources to loan us during the shuffle the
existing plan is still fine. We have not worked on future procurement at
all since our Foss4g conversation.

Thanks,
Alex

On 11/25/2014 11:58 AM, Justin Dugger via RT wrote:

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

neteler · December 4, 2014, 12:12pm

(looking forward to a faster projectsVM...)

ok, just a gentle ping to understand what's blocking here...

thanks and sorry for bothering
Markus

On Wed, Nov 26, 2014 at 12:45 AM, Alex Mandel
<tech_dev@wildintellect.com> wrote:

If you still have the resources to loan us during the shuffle the
existing plan is still fine. We have not worked on future procurement at
all since our Foss4g conversation.

Thanks,
Alex

On 11/25/2014 11:58 AM, Justin Dugger via RT wrote:

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

_______________________________________________
Sac mailing list
Sac@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/sac

Alex_Mandel1 · December 5, 2014, 11:00pm

OSUOSL got real busy, they contacted me last week to get the discussion
and scheduling going again. Will check again with them in a few days.

Thanks,
Alex

On 12/04/2014 04:12 AM, Markus Neteler wrote:

(looking forward to a faster projectsVM...)

ok, just a gentle ping to understand what's blocking here...

thanks and sorry for bothering
Markus

On Wed, Nov 26, 2014 at 12:45 AM, Alex Mandel
<tech_dev@wildintellect.com> wrote:

If you still have the resources to loan us during the shuffle the
existing plan is still fine. We have not worked on future procurement at
all since our Foss4g conversation.

Thanks,
Alex

On 11/25/2014 11:58 AM, Justin Dugger via RT wrote:

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

_______________________________________________
Sac mailing list
Sac@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/sac

--
Alex Mandel
http://wildintellect.com

neteler · December 11, 2014, 1:01pm

On Sat, Dec 6, 2014 at 12:00 AM, Alex Mandel <tech@wildintellect.com> wrote:

OSUOSL got real busy, they contacted me last week to get the discussion
and scheduling going again. Will check again with them in a few days.

Any news, Alex?

I/O wait on the projectVM is at time 48.1%wa

Thanks,

Markus

Alex_M · December 11, 2014, 5:42pm

Justin,

Any news, haven't heard back on this at all from your end?

Thanks,
Alex

On 11/25/2014 03:45 PM, Alex Mandel wrote:

If you still have the resources to loan us during the shuffle the
existing plan is still fine. We have not worked on future procurement at
all since our Foss4g conversation.

Thanks,
Alex

On 11/25/2014 11:58 AM, Justin Dugger via RT wrote:

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

_______________________________________________
Sac mailing list
Sac@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/sac

apache · December 11, 2014, 5:42pm

Justin,

Any news, haven't heard back on this at all from your end?

Thanks,
Alex

On 11/25/2014 03:45 PM, Alex Mandel wrote:

If you still have the resources to loan us during the shuffle the
existing plan is still fine. We have not worked on future procurement at
all since our Foss4g conversation.

Thanks,
Alex

On 11/25/2014 11:58 AM, Justin Dugger via RT wrote:

Hi all,

Sorry for the delays. With the school year starting up in October our large K-12 client really siphoned off a lot of time of ours dealing with Moodle upgrades and consequent performance issues. And in the process I lost track of some tickets.

Obviously, we failed to get this plan in place by October. And unfortunately, I'm caught up in meetings for much of the rest of the day, and the Thanksgiving Holiday means our staff won't be around much later this week, so we'll be missing November as well.

Does the original plan still make sense?

Justin

On Fri Sep 19 14:27:56 2014, tech@wildintellect.com wrote:

Justin & Team,

Based on our discussions at FOSS4G. Here's the plan to do some shuffling
and updates on the OSGeo hardware.

Overview - move VMs off osgeo4 and redo the OS and Raid (Software Raid 5)

Step 1.
Make sure Martin Spott knows how to use his new access of osgeo3 and
osgeo4 OS.

Step 2.
Change the mail.osgeo.osuosl.org VM to drbd
Sync it over to osgeo3, failover to osgeo
Change back to non-drbd

Step 3.
Schedule downtime (Mid-october) to turn off and copy Adhoc, Projects and
possibly QGIS to OSUOSL Ganeti cluster for a few days.
We might want a fresh Debian 7 VM for a QGIS rebuild - still checking

Step 4.
Upgrade/Redo the osgeo4 OS. OSUOSL has a new standard right? We're fine
with that as long as Martin gets access to troubleshoot.

Step 5.
Benchmark to make sure the disks are performing ok.

Step 6. Move Mail, QGIS, Adhoc and Projects back.

Long term, we will continue to investigate new hardware, docker, and are
open to suggestions on how to upgrade osgeo3 host OS with minimal
interruption.

Please get back to us with a schedule for who and when we might do each
of these steps.

Thanks,
Alex

_______________________________________________
Sac mailing list
Sac@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/sac

Alex_M · May 2, 2015, 6:52pm

On 03/06/2015 02:56 PM, Justin Dugger via RT wrote:

So we discussed this on IRC a few times since the last update on this ticket.

Here's the current plan I brought to the growth planning team:

Plan A
0. OSL sets aside 2U of space for OSGeo in a new rack.
1. OSGeo buys and ships 1U replacement hardware to the OSL to replace osgeo4
2. OSL sets up hardware in reserved space
3. (future) OSGeo buys and ships a second 1U server to replace osgeo3.
4. OSGeo migrates services on osgeo4 to new equipment.
5. OSL decomissions osgeo3/4

It looks like you've also requested another 2U of equipment, one for rackmount mac minis and another for a BSD build server. We've been decommissioning a lot of equipment lately, so we're approving the full 4U request.

We're looking at migrating Supercell from Ganeti to Openstack, so we're reluctant add new clients to the Ganeti cluster. But when that's prepared, you're welcome to join it.

When you're ready to ship things, let us know so we can start prepping on our end. Our shipping address can be found at http://osuosl.org/contact.

Justin we've placed the order for a 1U for item above. It will likely
ship next week from Silicon Mechanics. I will send you the tracking info
once I get it.

Once it arrives the plan is to let Martin do the setup remotely as was
done on our last machine.

Thanks,
Alex