Unable to reserve d760 and d760-hbm nodes

[Christine Guo]

Hi, 

I was wondering how to reserve the d760 and d760-hbm nodes from the Cloudlab Utah cluster? These nodes are listed on the hardware resource page but don't appear when I try to reserve them on the reservation request page.

Thank you so much for your help!

Best,

Christine

[Mike Hibler]

Those nodes are only for use via the Powder portal.

> --
> You received this message because you are subscribed to the Google Groups
> "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to cloudlab-user...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/
> 569a7b4c-1fa4-4f60-8284-510e31bfae9cn%40googlegroups.com.

[Mike Hibler]

Doh, I am wrong. We have too many node types that start with "d760"...

You can use the node types you mention, but you have to specify particular
nodes and not the type. Which of course you cannot do in a reservation.

Let me get back to you...

On Fri, Aug 22, 2025 at 08:03:32AM -0600, Mike Hibler wrote:
> Those nodes are only for use via the Powder portal.
>
> On Fri, Aug 22, 2025 at 06:10:11AM -0700, Christine Guo wrote:

> > Hi,??

> >
> > I was wondering how to reserve the d760 and d760-hbm nodes from the Cloudlab

> > Utah cluster? These nodes are listed on the hardware resource page??but don't

> > appear when I try to reserve them on the reservation request page.
> >
> > Thank you so much for your help!
> >
> > Best,
> > Christine
> >
> > --
> > You received this message because you are subscribed to the Google Groups
> > "cloudlab-users" group.
> > To unsubscribe from this group and stop receiving emails from it, send an email
> > to cloudlab-user...@googlegroups.com.
> > To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/
> > 569a7b4c-1fa4-4f60-8284-510e31bfae9cn%40googlegroups.com.
>
> --
> You received this message because you are subscribed to the Google Groups "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to cloudlab-user...@googlegroups.com.

> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/20250822140332.GA21701%40flux.utah.edu.

[Mike Hibler]

I should have just tried it. You can request these nodes by name in the
"Select Hardare" drop down for the Cloudlab Utah cluster. flex13-16 are the
d760 nodes, and flex11-12 are the d760-hbm nodes.

> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/20250822143658.GB21701%40flux.utah.edu.

[Christine Guo]

Hi Mike,

Thanks so much – I really appreciate it.

Best,

Christine

You received this message because you are subscribed to a topic in the Google Groups "cloudlab-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cloudlab-users/c9bBG36zScI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cloudlab-user...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/20250822150414.GC21701%40flux.utah.edu.

[Mohammad Tawhid Bhuiyan]

Hi,

I was able to successfully reserve a "flex11" node. However, when I try to create an instance using "flex11" as the physical node type on "select-hardware" profile, it fails with the following error:

*** WARNING: mapper: Improper type flex11 for node [vnode:node]! *** ERROR: mapper: *** Could not create vtop for [Experiment: ...-PG0/hbm] *** ERROR: mapper: 1 warnings.

I also tried using "d760-hbm" as physical node type, and got the same error.

Has anyone run into this issue before, or could you suggest how I might resolve it?

Thank you very much for your help!

Best regards,
Tawhid

[Mike Hibler]

"flex11" is not a node type, it is a node name. You have to reserve and
allocate "d760" and "d760-hbm" _types_ by the individual node _names_.
So you cannot use the select-hardware profile.

Try the specific-node profile instead:

https://www.cloudlab.us/p/PortalProfiles/specific-node

The URN for flex11 would be:

urn:publicid:IDN+utah.cloudlab.us+node+flex11

> cloudlab-users/20250822143658.GB21701%40flux.utah.edu.

>
> --
>
> You received this message because you are subscribed to a topic in the
> Google Groups "cloudlab-users" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic
> /cloudlab-users/c9bBG36zScI/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to
> cloudlab-user...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/
> cloudlab-users/20250822150414.GC21701%40flux.utah.edu.
>

> --
> You received this message because you are subscribed to the Google Groups
> "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to cloudlab-user...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/

> 0c391093-2b8a-4421-a830-f0a6e26d83c6n%40googlegroups.com.

[Mike Hibler]

In my earlier message when I said "You can request these nodes by name in
the "Select Hard(w)are" drop down, I meant if you are making a reservation
for these nodes through https://www.cloudlab.us/resgroup.php.

Sorry for the confusion...

On Wed, Aug 27, 2025 at 08:13:17PM -0600, Mike Hibler wrote:
> "flex11" is not a node type, it is a node name. You have to reserve and
> allocate "d760" and "d760-hbm" _types_ by the individual node _names_.
> So you cannot use the select-hardware profile.
>
> Try the specific-node profile instead:
>
> https://www.cloudlab.us/p/PortalProfiles/specific-node
>
> The URN for flex11 would be:
>
> urn:publicid:IDN+utah.cloudlab.us+node+flex11
>
> On Wed, Aug 27, 2025 at 06:39:39PM -0700, Mohammad Tawhid Bhuiyan wrote:
> > Hi,
> >
> > I was able to successfully reserve a "flex11" node. However, when I try to
> > create an instance using "flex11" as the physical node type on
> > "select-hardware" profile, it fails with the following error:
> > *** WARNING: mapper: Improper type flex11 for node [vnode:node]! *** ERROR:
> > mapper: *** Could not create vtop for [Experiment: ...-PG0/hbm] *** ERROR:
> > mapper: 1 warnings.
> >
> > I also tried using "d760-hbm" as physical node type, and got the same error.
> >
> > Has anyone run into this issue before, or could you suggest how I might resolve
> > it?
> >
> > Thank you very much for your help!
> >
> > Best regards,
> > Tawhid
> >
> > On Friday, August 22, 2025 at 5:33:27???PM UTC-4 Christine Guo wrote:
> >
> > Hi Mike,
> >

> > Thanks so much ??? I really appreciate it.

> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/20250828021317.GP21701%40flux.utah.edu.

[Mohammad Tawhid Bhuiyan]

Hi Mike,

Thanks a lot for your response! I reserved a "flex11" node, and then used "specific-node" profile with the URN as you shared. Now it shows the following error:

*** 1 nodes of type d760-hbm requested, but only 0 available nodes of type d760-hbm found

*** 1 nodes of type d760-hbm requested, but you are only allowed to use 0 Note that your topology cannot be instantiated on this cluster. You have most likely asked for hardware that does not exist, such as nodes of a type that do not exist, or more network interfaces that exist on any of the nodes at this cluster. You will need to modify your experiment or try a different cluster - re-submitting as-is will always result in failure!

Best,

Tawhid

[Mike Hibler]

We had an extra level of admission control applied during setup that I
forgot to remove. You should be able to allocate them now.

> 8ef53944-3ede-49f1-b99a-9d3c9e4df473n%40googlegroups.com.

[Mohammad Tawhid Bhuiyan]

Hi Mike,

It works now. Thank you so much!

Best,

Tawhid

[叶淼]

Hi Mike,

Thanks a lot for your response!

I have recently made a reservation for the "Flex16" node, but encountered the following error when starting the experiment:
No available physical nodes of type d760 found(1 requested)

Subsequently, the experiment indicated that a reservation exists, and will enter the pending state and retry periodically. However, the reservation was eventually terminated as it exceeded the 8-hour limit. 

Could you please help check this issue?

best,f

Miao

[Christine Guo]

Hi Mike, 

I tried to reserve the d760 nodes (no longer named flex13-16) and it says that Admission Control has limited me to 0 nodes. I was wondering if this is an error and if I could get access to the machine for bare-metal access? 

Thank you!

Best,

Christine

To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/bc6e7a81-401f-4eac-83d9-fa7585310a37n%40googlegroups.com.

[Mike Hibler]

We have changed the access model for the "flex" nodes. Previously, you
needed to reserve and allocate a specific node (as you have been doing).
Now you can allocate them by type (e.g., "d760") but your project has to
be approved to use the type. While there is no need to reserve them first,
we still recommend that you do so since they are popular nodes. You can
also still allocate a specific flex node, but you cannot reserve a specific
node.

I have enabled your project for the "d760" and "d760-hbm" types. We would
prefer you not use the latter unless you need the high bandwidth memory.

> cloudlab-users/bc6e7a81-401f-4eac-83d9-fa7585310a37n%40googlegroups.com.

>
> --
> You received this message because you are subscribed to the Google Groups
> "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to cloudlab-user...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/

> CAG1%2BtBSa-ur1zrucBcbVmZSRhu7%3DtkPdxBOa2v1KOYkGQU1fiQ%40mail.gmail.com.

[Mike Hibler]

BTW, these new "Flex" nodes (d7615, d760, d760-hbm) are now documented in
the Manual:
https://docs.cloudlab.us/hardware.html#%28part._cloudlab-utah%29

On Wed, Nov 12, 2025 at 03:50:30PM -0700, Mike Hibler wrote:
> We have changed the access model for the "flex" nodes. Previously, you
> needed to reserve and allocate a specific node (as you have been doing).
> Now you can allocate them by type (e.g., "d760") but your project has to
> be approved to use the type. While there is no need to reserve them first,
> we still recommend that you do so since they are popular nodes. You can
> also still allocate a specific flex node, but you cannot reserve a specific
> node.
>
> I have enabled your project for the "d760" and "d760-hbm" types. We would
> prefer you not use the latter unless you need the high bandwidth memory.
>
> On Wed, Nov 12, 2025 at 05:15:55PM -0500, Christine Guo wrote:

> > Hi Mike,??

> >
> > I tried to reserve the d760 nodes (no longer named flex13-16) and it says that
> > Admission Control has limited me to 0 nodes. I was wondering if this is an

> > error and if I could get access to the machine for bare-metal access???

> >
> > Thank you!
> >
> > Best,
> > Christine
> >
> > On Fri, Sep 5, 2025 at 11:38???PM ?????? <yem...@gmail.com> wrote:
> >
> > Hi Mike,
> > Thanks a lot for your response!
> > I have recently made a reservation for the "Flex16" node, but encountered
> > the following error when starting the experiment:
> > No available physical nodes of type d760 found(1 requested)
> > Subsequently, the experiment indicated that a reservation exists, and will
> > enter the pending state and retry periodically. However, the reservation

> > was eventually terminated as it exceeded the 8-hour limit.??

> > Could you please help check this issue?
> >
> > best,f
> > Miao
> >

> > ???2025???8???29???????????? UTC+8 06:29:13<Mike Hibler> ?????????

> To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/20251112225030.GC1107%40flux.utah.edu.

[Christine Guo]

Hi Mike,

Thank you so much for your help! I really appreciate it.

Best,

Christine

To view this discussion visit https://groups.google.com/d/msgid/cloudlab-users/20251112225548.GD1107%40flux.utah.edu.

[Christine Guo]

Hi Mike,

I hope you're doing well!

I currently have a reservation for a d760 node that has just started but it seems that when trying to start an experiment for any of the flex13-16, I'm getting "Internal error creating experiment." Do you know what may be causing this problem? 

Thank you!

Best,

Christine

[ajma...@gmail.com]

Hi Christine,

We had a cert issue pop up this morning which is preventing people from starting experiments.  We're working on it and will let you know when it's resolved.  Sorry for the inconvenience.

Best,

 - Aleks