On 04/09/14 at 01:05pm, Steve Dickson wrote:
On 04/08/2014 10:56 PM, Baoquan He wrote:
> On 04/08/14 at 08:41am, Steve Dickson wrote:
>> On 04/07/2014 10:35 PM, Baoquan He wrote:
>>> Thanks, Steve and Karel
>>> So now in nfs file system, nfsv1/2/3 use the same file_system_type,
>>> namely &nfs_fs_type. However in nfsv4, it uses its onw file_system_type
>>> like below. I am gonna to ask if this can be changed back to be
>>> just like the previous 3 version.
>> The issue here is the fact the NFS versions were broken out into separate
>> kernel modules in RHEL7. This is actually the reason we could turn
>> off v2 support... So the option of changing it back will not work
>> since two different modules can have the same name.
>> I would suggest you adapt you script to handle both....
> Hi Steve,
> Yeah, I tried changing kernel code, then mount.nfs failed immediately. I
> will adapt my script as you suggested. But I will create a bug to track
> this issue. Maybe whatever the version of nfs is, the type should be
> nfs, people can check its version from the mount option, and that truly
> has a option to show version, namely v3 or v4.
Well there is not much we can do about what is being put
into /proc/self/mountinfo... But... Looking at what is being
put /proc/self/mountinfo.... nfs == v3 and nfs4 == v4
Those are the *only* two entries... So I don't see why a
simple change to your script like:
if [ "$_fstype" = "nfs" -o "$_fstype" = "nfs4"
would not fix the problem?
Yeah, it's easy to adapt kdump script. But I don't think only kdump
encounter this problem. This is more likely a generic issue, why fstype
of nfs3 is nfs, nfs4 is nfs4, why user have to know this is because
kernel separate a implementation for nfs4.
Anyway I will make change in kdump script, also a bug for this issue has
been created, its priority is set very low.