On Sat, Sep 03 2016, james harvey wrote:
>
> FWIW, after nfs-server eventually starts, systemctl still shows
> rpcbind.socket (and rpcbind.service) as masked, inactive, and dead.
I think this is a kernel bug. You can fix it by reverting
Commit: 4b0ab51db32e ("SUNRPC: xs_sock_mark_closed() does not need to trigger socket autoclose")
See also http://www.spinics.net/lists/linux-nfs/msg59264.html and
follow-ups.
>>>
>>> # rpc.mountd --foreground -N 2 -N 3
>>>
>>> {{{ another tty }}}
>>> # strace rpc.nfsd --debug -N 2 -N 3 8
>>>
>>> Gives the attached strace output, which shows the same two really long
>>> delays when reading or writing /proc/fs/nfsd/portlist.
>>>
>> I don't see the attachment...
>>
>> steved.
>
> Strange, wonder if the list removed it - I do see it on my sent message.
>
> Anyways, I just uploaded it here: http://pastebin.ca/3709748
>
> The delays happen at lines 145-149, which print as one line, I just
> broke it apart showing the characters that show on the line before and
> after the delay.
The delay at 'read' is waiting for nfsd_mutex. Probably something else
was trying to start up nfsd and hit the 3 minute wait while writing the
threads. Maybe you interrupted that or backgrounded it or something.
>
> Same at lines 166-170.
This one is the real problem. Starting nfsd tries to register with
rpcbind (even for v4-only) and (due to the bug) it takes 3 minutes to
timeout.
NeilBrown
> --
> To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Attachment:
signature.asc
Description: PGP signature
