[lustre-discuss] Lustre Module ko2iblnd does not load for Lustre 2.16.1

[Pati, Abhilasha]

Greetings To All,


I am trying to install the lustre server for Lustre version 2.16.1 on RHEL 9.4 ....and the command I am using for it is dnf install lustre-all-dkms.

Unfortunately....I get the following messages from kernel buffer...


[167702.252133] in_kernel_ko2iblnd: Unknown symbol rdma_set_reuseaddr (err -22)
[167702.252135] in_kernel_ko2iblnd: disagrees about version of symbol ib_destroy_cq_user
[167702.252135] in_kernel_ko2iblnd: Unknown symbol ib_destroy_cq_user (err -22)
[167702.252138] in_kernel_ko2iblnd: disagrees about version of symbol ib_modify_qp
[167702.252138] in_kernel_ko2iblnd: Unknown symbol ib_modify_qp (err -22)
[167702.252144] in_kernel_ko2iblnd: disagrees about version of symbol ib_dma_virt_map_sg
[167702.252145] in_kernel_ko2iblnd: Unknown symbol ib_dma_virt_map_sg (err -22)
[167702.252146] in_kernel_ko2iblnd: disagrees about version of symbol rdma_destroy_id
[167702.252147] in_kernel_ko2iblnd: Unknown symbol rdma_destroy_id (err -22)
[167702.252152] in_kernel_ko2iblnd: disagrees about version of symbol rdma_accept
[167702.252152] in_kernel_ko2iblnd: Unknown symbol rdma_accept (err -22)
[167702.252160] in_kernel_ko2iblnd: disagrees about version of symbol ib_dealloc_pd_user
[167702.252160] in_kernel_ko2iblnd: Unknown symbol ib_dealloc_pd_user (err -22)

followed by the following LnetError and the LustreError:

[167702.282546] LNetError: 8805:0:(api-ni.c:2616:lnet_load_lnd()) Can't load LND o2ib, module ko2iblnd, rc=256
[167702.282713] LustreError: 8805:0:(events.c:640:ptlrpc_init_portals()) network initialisation failed: rc = -22

I must add to this that we have installed the mellanox diver on the machine as well...which has the following vermagic 5.14.0-427.31.1_lustre.el9.x86_64, which is the same as our kernel version...Infact the ib_core module on which this ko2iblnd depends...also has the vermagic : 5.14.0-427.31.1_lustre.el9.x86_64.

Not sure if this is related to the above issue , but during the installation of lustre, we come across an error message that says that the build of lquota.ko module fails.

Error! Build of lustre/lquota/lquota.ko failed for: 5.14.0-427.31.1_lustre.el9.x86_64 (x86_64) Make sure the name and location of the generated module are correct, or consult /var/lib/dkms/lustre-all/2.16.1/build/make.log for more information. warning: %post(lustre-all-dkms-2.16.1-1.el9.noarch) scriptlet failed, exit status 7

and when the log file is checked.....the following is returned (not sure if this is an error or a warning):

####### cat /var/lib/dkms/lustre-all/2.16.1/build/make.log | grep quota
checking if 'quotactl_ops.set_dqblk' takes struct qc_dqblk... yes
checking whether to enable quota support global control... yes
config.status: creating lustre/quota/Makefile
config.status: creating lustre/quota/autoMakefile
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lproc_quota.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota_lib.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota_disk.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota_entry.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_request.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_lib.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_entry.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_lock.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_reint.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/osd-zfs/osd_quota.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_writeback.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_config.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qsd_handler.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qmt_dev.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/osc/osc_quota.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qmt_handler.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qmt_lock.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qmt_entry.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/qmt_pool.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/osd-ldiskfs/osd_quota.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/osd-ldiskfs/osd_quota_fmt.o
  LD [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota.o
  CC [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota.mod.o
  LD [M]  /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota.ko
  BTF [M] /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota.ko
Skipping BTF generation for /var/lib/dkms/lustre-all/2.16.1/build/lustre/quota/lquota.ko due to unavailability of vmlinux
Making all in quota
make[3]: Entering directory '/var/lib/dkms/lustre-all/2.16.1/build/lustre/quota'
make[3]: Leaving directory '/var/lib/dkms/lustre-all/2.16.1/build/lustre/quota'

We would really appreciate if some insight was given into why this error occurs and how it could be resolved.

With sincerest regards,
Abhilasha Pati
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20250916/3316cce6/attachment.htm>