| 1 |
"Brian Budge" writes: |
| 2 |
> Hi Bryan - |
| 3 |
> |
| 4 |
> I don't seem to have a 40-ib.rules in any of my /etc/udev/rules.d on any |
| 5 |
> node. |
| 6 |
|
| 7 |
Aha. That file is part of sys-cluster/openib-drivers, which you don't have |
| 8 |
installed. You can use the infiniband support that is part of the kernel, |
| 9 |
but the driver versions won't match your openib userspace software versions |
| 10 |
(not necessarily a problem), and you'll be missing the startup scripts. |
| 11 |
|
| 12 |
Older versions of the files are installed by the openib-files ebuild. That |
| 13 |
ebuild is currently incompatible with openib-userspace, though it perhaps |
| 14 |
shouldn't be. I guess I could fix that so you could install openib-files. |
| 15 |
But if possible, I'd recommend turning off the kernel builtin drivers, and |
| 16 |
emerge openib-drivers. |
| 17 |
|
| 18 |
In the mean time, just installing '40-ib.rules' might help, but I'm not |
| 19 |
sure. |
| 20 |
|
| 21 |
#### /etc/udev/rules.d/40-ib.rules #### |
| 22 |
KERNEL=="umad*", NAME="infiniband/%k" |
| 23 |
KERNEL=="issm*", NAME="infiniband/%k" |
| 24 |
KERNEL=="ucm*", NAME="infiniband/%k", MODE="0666" |
| 25 |
KERNEL=="uverbs*", NAME="infiniband/%k", MODE="0666" |
| 26 |
KERNEL=="uat", NAME="infiniband/%k", MODE="0666" |
| 27 |
KERNEL=="ucma", NAME="infiniband/%k", MODE="0666" |
| 28 |
KERNEL=="rdma_cm", NAME="infiniband/%k", MODE="0666" |
| 29 |
######## |
| 30 |
|
| 31 |
> My /sys/class/infiniband directory contains mthca0, which contains: |
| 32 |
> > ls -la /sys/class/infiniband/mthca0/ |
| 33 |
> total 0 |
| 34 |
> drwxr-xr-x 3 root root 0 Jan 2 20:54 . |
| 35 |
> drwxr-xr-x 3 root root 0 Jan 2 20:54 .. |
| 36 |
> -r--r--r-- 1 root root 4096 Jan 2 21:07 board_id |
| 37 |
> lrwxrwxrwx 1 root root 0 Jan 3 00:01 device -> |
| 38 |
> ../../../devices/pci0000:20/0000:20:0a.0/0000:21:00.0 |
| 39 |
> -r--r--r-- 1 root root 4096 Jan 2 21:07 fw_ver |
| 40 |
> -r--r--r-- 1 root root 4096 Jan 2 21:07 hca_type |
| 41 |
> -r--r--r-- 1 root root 4096 Jan 2 21:07 hw_rev |
| 42 |
> -rw-r--r-- 1 root root 4096 Jan 2 21:07 node_desc |
| 43 |
> -r--r--r-- 1 root root 4096 Jan 2 21:07 node_guid |
| 44 |
> -r--r--r-- 1 root root 4096 Jan 2 21:06 node_type |
| 45 |
> drwxr-xr-x 3 root root 0 Jan 2 21:07 ports |
| 46 |
> lrwxrwxrwx 1 root root 0 Jan 3 00:01 subsystem -> |
| 47 |
> ../../../class/infiniband |
| 48 |
> -r--r--r-- 1 root root 4096 Jan 2 21:07 sys_image_guid |
| 49 |
> --w------- 1 root root 4096 Jan 2 20:54 uevent |
| 50 |
|
| 51 |
Is this what is shown on the node that does not have '/dev/infiniband'? |
| 52 |
|
| 53 |
What about '/sys/class/infiniband_verbs/'? |
| 54 |
|
| 55 |
-- |
| 56 |
gentoo-cluster@g.o mailing list |
Archives