Unable to start the Client service

[Info Technologies]

Hi Team

I have 1 node and 2 disks .

Cluster was running fine when I did followed the steps but after installing Nvidia GPU drivers , Needed to reboot the server . After reboot mounts were gone , so I deleted all the mount data , all the directories and uninstall beegfs using yum remove .

Now I started everything again from fresh and see below issue when starting client service .

Please refer to attachment

Please let me know how to proceed .

[Info Technologies]

Adding status for services

please refer attachment

[Eric Weber]

The output of "yum -y install kernel-devel" in your second attachment includes indicates that two versions of kernel-devel are currently installed. 4.18.0-147 is the version of the kernel that shipped with RHEL/CentOS 8.1. 4.18.0-348 is the version of the kernel that shipped with RHEL/CentOS 8.5. See https://access.redhat.com/articles/3078 for details. I am guessing you inadvertently upgraded from RHEL/CentOS 8.1 to RHEL/CentOS 8.5 when installing the NVIDIA GPU drivers and are now running on the 8.5 kernel. You can confirm using "cat /etc/redhat-release" or "uname -r". Unfortunately, there are known issues building the BeeGFS client module for the 8.5 kernel. 

As for what to do now, this is a somewhat complicated question. If you have support through ThinkParQ or another vendor, I recommend opening a ticket right away. If you are running without support, see any number of posts in this user group for similar symptoms and discussions (e.g. https://groups.google.com/g/fhgfs-user/c/XVspSRtcFFQ). The answers seem to boil down to one of the following:

• Attempt to roll back/downgrade the kernel.
• Patch BeeGFS yourself.
• Wait for an official patch from ThinkParQ.

All of these options have potential downsides, so please proceed with caution.

Eric Weber

Software Engineer

E/EF Series Solutions

NetApp

[Info Technologies]

Thanks Eric for the update .

Right now I am Cent OS 8.5 ,  able to start beegfs on single node (Refer ScreenShot) .

Issue is coming now for enabling BeeOND on the same node . (Refer Sceenshot) .

I have tried below options for fixing the issue but beegfs meta server goes for a six .

1. storeUseExtendedAttribs      = true --> false

2. sysMountSanityCheckMS         = 11000 --> 0

Regards

Karan Singh

[Info Technologies]

++ Adding 

Checked with 8.1 CentOS , same issue with Beeond

[Info Technologies]

++ Adding 

Changed the below parameters but Beegfs of 6.9 T is not showing , so it means that client can only use that space (3.5T) which has been assigned by Beeond mount point ??

[Info Technologies]

++Adding 

Remove all the above components and made 2 servers with just (yum install beeond)

Getting below issue when starting beeond .

(2) 19:07:51 *mount(6923) [Remoting (stat storage targets)] >> Error target (storage): 2; Msg: Unknown storage target
(0) 19:07:51 *mount(6923) [Mount sanity check] >> Retrieval of storage server free space info failed. Are the storage servers running and registered at the management daemon? Did you remove a storage target directory on a server? (Error: Unknown storage target)
(2) 19:07:51 *mount(6923) [App (stop components)] >> Stopping components...

Please let me know , what wrong i am doing ???

Config details below in screenshot

[Eric Weber]

I'm not sure what's behind this new BeeOND issue, but if I understand you correctly, you were able to successfully circumvent your previous issue and get the BeeGFS 7.2.5 client up and running on CentOS 8.5? Would you mind sharing what you did to make that work (at least at a high level)? CentOS 8.5 issues come up a lot in this user group and I think the community would benefit from a brief explanation.

Eric Weber

Software Engineer

E/EF Series Solutions

NetApp