Failed to run instance with flavor set with GPU Passthrough, Exceeded max scheduling attempts 3 for instance a3927c7e-cfce-4f2d-bf0f-64ffabd449a6. Last exception: XML error: Hostdev already exists in the domain configuration
Issue
-
We have enable pci passthrough for gpu, we use tesla t4 .
-
During process creation we facing error like this:
Exceeded maximum number of retries. Exceeded max scheduling attempts 3 for instance a3927c7e-cfce-4f2d-bf0f-64ffabd449a6. Last exception: XML error: Hostdev already exists in the domain configuration
- This is the flavor being used:
[stack@director ~]$ openstack flavor show a05ab4e0-83e4-4417-aa53-f60480eaa691
+----------------------------+--------------------------------------+
| Field | Value |
+----------------------------+--------------------------------------+
| OS-FLV-DISABLED:disabled | False |
| OS-FLV-EXT-DATA:ephemeral | 0 |
| access_project_ids | None |
| description | None |
| disk | 200 |
| id | a05ab4e0-83e4-4417-aa53-f60480eaa691 |
| name | C16M64D200-GPU |
| os-flavor-access:is_public | True |
| properties | pci_passthrough:alias='t4:1' |
| ram | 65536 |
| rxtx_factor | 1.0 |
| swap | |
| vcpus | 16 |
+----------------------------+--------------------------------------+
- Here's the nova.conf for pci passthrough/alias:
$ cat nova.conf | grep alias
alias={"device_type":"type-PF","name":"t4","product_id":"1eb8","vendor_id":"10de"}
$ cat nova.conf | grep whitelist
passthrough_whitelist={"product_id":"1eb8","vendor_id":"10de"}
lspci
from the compute:
$ grep -i nvidia lspci
5e:00.0 3D controller [0302]: NVIDIA Corporation TU104GL [Tesla T4] [10de:1eb8] (rev a1)
Environment
- Red Hat OpenStack Platform 17.1 (RHOSP)
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.