GPU instances failing to be created

10,726 views
Skip to first unread message

Motiur Rahman

unread,
Nov 8, 2017, 8:50:44 AM11/8/17
to gce-discussion
Hi, hello - I am trying to create instance of 1 GPU K80 with 4 cpus , 26 GB RAM from us-west-1b region . I am getting this odd error - 

The request contains invalid arguments: "Instances with guest accelerators do not support live migration.". Error code: "badRequest"

Is this normal, or is it going to persist. I am trying to running some deep learning codes using the GPU. Any suggestion will be helpful.

Thanks
Mohammad

Digil (Google Cloud Platform Support)

unread,
Nov 8, 2017, 11:50:34 AM11/8/17
to gce-discussion

I tried to create a similar instance and saw the same error. It seems that you have to request a quota on GPU. Make sure that you have the necessary quota for using the GPU on an instance.


There is this help center article that will help you to create the instance with GPUs. Also, do check the restrictions mentioned in the documentation on creating the GPU integrated instances.


It is also clearly mentioned here in the introductory part that GPU instance has to terminate for host maintenance events and thus cannot live migrate.

Motiur Rahman

unread,
Nov 8, 2017, 1:05:44 PM11/8/17
to gce-discussion
I had quota issues before. I believe I have resolved it. This new problem that I am experiencing has something to do with maintenance issue.
If that's the case do you know when will it be resolved.

Thanks. 

Michael Basilyan

unread,
Nov 8, 2017, 1:09:12 PM11/8/17
to Motiur Rahman, gce-discussion, Chris Kleban, Ari Garcia
This is likely because of this:

When you create an instance with one or more GPUs, you must set the instance to terminate on host maintenance. Instances with GPUs cannot live migrate because they are assigned to specific hardware devices. See GPU restrictions for details.


So for example:

gcloud compute instances create [INSTANCE_NAME] \
   
--machine-type [MACHINE_TYPE] --zone [ZONE] \
   
--accelerator type=[ACCELERATOR_TYPE],count=[ACCELERATOR_COUNT] \
   
--image-family [IMAGE_FAMILY] --image-project [IMAGE_PROJECT] \
   
--maintenance-policy TERMINATE --restart-on-failure \
   
--metadata startup-script='[STARTUP_SCRIPT]'

--
© 2017 Google Inc. 1600 Amphitheatre Parkway, Mountain View, CA 94043
 
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-discussion@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.
---
You received this message because you are subscribed to the Google Groups "gce-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gce-discussion+unsubscribe@googlegroups.com.
To post to this group, send email to gce-discussion@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gce-discussion/0839f8ac-0dea-4937-8143-c754df02fb76%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

Message has been deleted

David Knight

unread,
Nov 26, 2017, 2:34:06 PM11/26/17
to gce-discussion
If you are creating an instance via the Web interface, find "Availability Policy"->"Automatic Restart" and set it to "Off"
Email preferences: You received this email because you signed up for the Google Compute Engine Discussion Google Group (gce-dis...@googlegroups.com) to participate in discussions with other members of the Google Compute Engine community and the Google Compute Engine Team.

---
You received this message because you are subscribed to the Google Groups "gce-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gce-discussio...@googlegroups.com.
To post to this group, send email to gce-dis...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages