Question about vfiler failover

[NATHANKIPPEN]

Question.

I'm being told my netapp support that if I have nfsv4 option turned on for one vfiler (not vfiler0) then when that vfiler fails over to the other filer (head), the nfsv4 option will not be turned on, unless vfiler0 on the head that tookover had nfsv4 enabled is that true?  Can somebody point me to the documentation on that?

Thanks,

[scottgelb]

Have you tested this?  I would ask support to escalate for more information.  NFS v4 is set in options and the vFiler itself has the option... global settings in vfiler0 like IPspaces and partner interfaces absolutely need to be on the partner, but I don't see why protocol settings need to be in both when already set in the vFiler itself.

For example..disabled in vfiler0 but enabled in vfiler1.

vsim-7m-1> vfiler run * options nfs.v4

===== vfiler0

nfs.v4.acl.enable            off

nfs.v4.enable                off

nfs.v4.id.allow_numerics     on

nfs.v4.id.domain

nfs.v4.read_delegation       off

nfs.v4.write_delegation      off

===== vfiler1

nfs.v4.acl.enable            off

nfs.v4.enable                on

nfs.v4.id.allow_numerics     on

nfs.v4.id.domain             lab.local

nfs.v4.read_delegation       off

nfs.v4.write_delegation      off

[NATHANKIPPEN]

I haven't had the chance to test this yet..  However there is some evidence that this information could be true:

filer02> options nfs.v4 nfs.v4.acl.enable                 off        (value might be overwritten in takeover)

nfs.v4.enable                                                       off     (value might be overwritten in takeover)

nfs.v4.id.domain nfs.v4.read_delegation                 off        (value might be overwritten in takeover)

nfs.v4.write_delegation                                         off        (value might be overwritten in takeover)

If vfiler0, on the other head, takes over control of the failed vfiler, looks like some options will be overwritten.

[scottgelb · Accepted solution]

In 7-mode a virtual instance is taken over of vfiler0, so certain settings like autosupport, timed, nfs per below do matter…but individual vFilers that have these options already set run in the virtual instance with the same settings regardless of the node they run on… autosupport is vfiler0 only so would need to be the same… but v4 enable is in the vFiler and can be different than vfiler0 so I don’t see the dependency based on this for vFilers… for data in vfiler0 it would though.

[NATHANKIPPEN]

I agree.  The options in vfiler0 need to be the same on both heads, but for example the NFS options of a vfiler (excluding vfiler0) should stay with that vfiler on failover.  I have asked NetApp to investigate this further and provide supporting documentation.  I'll let you know when I have the opportunity to test or if I get some official documentation from netapp.  Thank you for the responses.

[NATHANKIPPEN]

After speaking with NetApp again they said the following:

"I apologize for the confusion as I did not catch the filer names in the original email so that is why we have had a disconnect.   You are correct in your statements; vfilers other than vfiler0 are independent and will be failed over with their options in the event of a CFO.  However as I stated and you stated below vfiler0 must be configured as a pair and have identical options to failover correctly.  Again I apologize for the confusion as I did make a mistake on the vfiler naming ..."

Thanks again for your replies Scott.

[scottgelb]

Very good… support followed up well.