Re: NFS 4.1 RECLAIM_COMPLETE FS failed error in combination with ESXi client

[Rick Macklem]

NAGY Andreas wrote:
>attached the trace. If I see it correct it uses FORE_OR_BOTH. (bctsa_dir: >CDFC4_FORE_OR_BOTH (0x00000003))
Yes. The scary part is the ExchangeID before the BindConnectiontoSession.
(Normally that is only done at the beginning of a new mount to get a ClientID,
 followed immediately by a CreateSession. I don't know why it would do this?)

The attached patch might get BindConnectiontoSession to work. I have no way
to test it beyond seeing it compile. Hopefully it will apply cleanly.

>The trace is only with the first patch, have not compiled the wantdeleg patches so >far.
That's fine. I don't think that matters much.

>I think this is related to the BIND_CONN_TO_SESSION; after a disconnect the ESXi >cannot connect to the NFS also with this warning:
>2018-03-07T16:55:11.227Z cpu21:66484)WARNING: NFS41: NFS41_Bug:2361: >BUG - Invalid BIND_CONN_TO_SESSION error: NFS4ERR_NOTSUPP
If the attached patch works, you'll find out what it fixes.

>Another thing I noticed today is that it is not possible to delete a folder with the >ESXi datastorebrowser on the NFS mount. Maybe it is a VMWare bug, but with >NFS3 it works.
>
>Here the vmkernel.log with only one connection contains mounting, trying to >delete a folder and disconnect:
>
>2018-03-07T16:46:04.543Z cpu12:68008 opID=55bea165)World: 12235: VC opID >c55dbe59 maps to vmkernel opID 55bea165
>2018-03-07T16:46:04.543Z cpu12:68008 opID=55bea165)NFS41: >NFS41_VSIMountSet:423: Mount server: 10.0.0.225, port: 2049, path: /, label: >nfsds1, security: 1 user: , options: <none>
>2018-03-07T16:46:04.543Z cpu12:68008 opID=55bea165)StorageApdHandler: >977: APD Handle  Created with lock[StorageApd-0x43046e4c6d70]
>2018-03-07T16:46:04.544Z cpu11:66486)NFS41: >NFS41ProcessClusterProbeResult:3873: Reclaiming state, cluster 0x43046e4c7ee0 >[7]
>2018-03-07T16:46:04.545Z cpu12:68008 opID=55bea165)NFS41: >NFS41FSCompleteMount:3791: Lease time: 120
>2018-03-07T16:46:04.545Z cpu12:68008 opID=55bea165)NFS41: >NFS41FSCompleteMount:3792: Max read xfer size: 0x20000
>2018-03-07T16:46:04.545Z cpu12:68008 opID=55bea165)NFS41: >NFS41FSCompleteMount:3793: Max write xfer size: 0x20000
>2018-03-07T16:46:04.545Z cpu12:68008 opID=55bea165)NFS41: >NFS41FSCompleteMount:3794: Max file size: 0x800000000000
>2018-03-07T16:46:04.545Z cpu12:68008 opID=55bea165)NFS41: >NFS41FSCompleteMount:3795: Max file name: 255
>2018-03-07T16:46:04.545Z cpu12:68008 opID=55bea165)WARNING: NFS41: >NFS41FSCompleteMount:3800: The max file name size (255) of file system is >larger than that of FSS (128)
>2018-03-07T16:46:04.546Z cpu12:68008 opID=55bea165)NFS41: >NFS41FSAPDNotify:5960: Restored connection to the server 10.0.0.225 mount >point nfsds1, mounted as 1a7893c8-eec764a7-0000-000000000000 ("/")
>2018-03-07T16:46:04.546Z cpu12:68008 opID=55bea165)NFS41: >NFS41_VSIMountSet:435: nfsds1 mounted successfully
>2018-03-07T16:47:19.869Z cpu21:67981 opID=e47706ec)World: 12235: VC opID >c55dbe91 maps to vmkernel opID e47706ec
>2018-03-07T16:47:19.869Z cpu21:67981 opID=e47706ec)WARNING: NFS41: >NFS41FileOpReaddir:4728: Failed to process READDIR result for fh 0x43046e4c6
I have no idea if getting BindConnectiontoSession working will fix this or not?

rick

-------------- next part --------------
A non-text attachment was scrubbed...
Name: bindconn.patch
Type: application/octet-stream
Size: 5343 bytes
Desc: bindconn.patch
URL: <http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20180308/320d504a/attachment.obj>