Recently we got some new KNL nodes and decided to try RHEL 7.3 on these hosts running the 3.10.0-514.10.2.el7.x86_64 kernel.
After installing the IntelOPA-Basic software and upgrading the firmware on the HFI and rebooting the nodes, we still can't get anything other than the following from opainfo
[root@sknl0701 ~]# opainfo
oib_utils ERROR: [7534] open_verbs_ctx: failed to find verbs device
opainfo: Unable to open hfi:port 0:1
Even though the software and firmware never complains about any errors we can still see that even after forcing dracut to recreate the system image the hfi1 driver will not load.
[root@shas0101 ~]# lsmod | grep hfi1
hfi1 633634 1
rdmavt 57992 1 hfi1
ib_mad 47817 5 hfi1,ib_cm,ib_sa,rdmavt,ib_umad
ib_core 98787 14 hfi1,rdma_cm,ib_cm,ib_sa,iw_cm,xprtrdma,ib_mad,ib_ucm,rdmavt,ib_iser,ib_umad,ib_uverbs,ib_ipoib,ib_isert
i2c_algo_bit 13413 2 hfi1,mgag200
i2c_core 40582 6 drm,hfi1,ipmi_ssif,drm_kms_helper,mgag200,i2c_algo_bit
[root@sknl0701 ~]# modprobe -v hfi1
[root@sknl0701 ~]# lsmod | grep hfi1
hfi1 697628 0
rdmavt 63294 1 hfi1
ib_core 210381 13 hfi1,rdma_cm,ib_cm,iw_cm,rpcrdma,ib_ucm,rdmavt,ib_iser,ib_umad,ib_uverbs,rdma_ucm,ib_ipoib,ib_isert
i2c_algo_bit 13413 3 igb,hfi1,mgag200
i2c_core 40756 7 drm,igb,hfi1,ipmi_ssif,drm_kms_helper,mgag200,i2c_algo_bit
[root@sknl0701 ~]# yum info libibmad
Loaded plugins: product-id, search-disabled-repos, subscription-manager
Installed Packages
Name : libibmad
Arch : x86_64
Version : 1.3.12
Release : 1.el7
Size : 132 k
Repo : installed
From repo : curc
Summary : OpenFabrics Alliance InfiniBand MAD library
URL : http://openfabrics.org/
License : GPLv2 or BSD
Description : libibmad provides low layer IB functions for use by the IB diagnostic
: and management programs. These include MAD, SA, SMP, and other basic
: IB functions.
[root@sknl0701 ~]# yum info libibmad-devel
Loaded plugins: product-id, search-disabled-repos, subscription-manager
Installed Packages
Name : libibmad-devel
Arch : x86_64
Version : 1.3.12
Release : 1.el7
Size : 50 k
Repo : installed
From repo : curc
Summary : Development files for the libibmad library
URL : http://openfabrics.org/
License : GPLv2 or BSD
Description : Development files for the libibmad library.
libibmad was and has been installed on the new node as well, so I am out of ideas at the moment. Any help would be appreciated!
Clik here to view.