Bug 817041 - rpc.idmapd not started on client
Summary: rpc.idmapd not started on client
Alias: None
Product: Fedora
Classification: Fedora
Component: nfs-utils
Version: 17
Hardware: Unspecified
OS: Unspecified
Target Milestone: ---
Assignee: Steve Dickson
QA Contact: Fedora Extras Quality Assurance
Depends On:
TreeView+ depends on / blocked
Reported: 2012-04-27 13:50 UTC by Richard W.M. Jones
Modified: 2012-05-17 20:43 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Last Closed: 2012-05-17 20:43:02 UTC
Type: Bug

Attachments (Terms of Use)

Description Richard W.M. Jones 2012-04-27 13:50:07 UTC
Description of problem:

rpc.idmapd is required to get nfsv4 to work usefully.  It should
be running on *both* the client and server.

However on Fedora 17 it is not started on the client side.  I
had to write an rc.local script to start it by hand:

#!/bin/bash -
# Fix NFS.
sleep 10
systemctl start var-lib-nfs-rpc_pipefs.mount


(1) sleep 10 is needed because rc.local runs too early.
(2) /var/lib/nfs/rpc_pipefs is not mounted even though the
mountpoint and service exists.  No idea why.

I tried setting NEED_IDMAPD=yes in /etc/default/nfs and
/etc/default/nfs-common, but neither made any difference.

Version-Release number of selected component (if applicable):


Comment 1 Matt Kinni 2012-05-09 23:30:58 UTC
The workaround I've been using is
systemctl enable nfs-server.service on the client.

I only need the client to be the "client" but that service starts idmapd so meh

Comment 2 Steve Dickson 2012-05-12 11:37:13 UTC
What kernel version are you using?

With 3.3 and beyond the idmapped was changed to use a keyring based idmapping using the nfsidmap command. See nfsidmap(5) for details. With the being the cause the rpc.idmapd daemon no longer needed to be started on the NFS client.

Comment 3 Richard W.M. Jones 2012-05-12 12:22:37 UTC
Interesting ...

This is arm, so it was:
Linux trim.home.annexia.org #1 SMP PREEMPT Fri Jan 20 11:31:29 EST 2012 armv7l armv7l armv7l GNU/Linux
because there *was* a problem booting this hardware with the
newer kernel.  Since (even) newer kernels are available I need
to try those.

I'll try to reboot this machine later today (currently it's running
some crucial builds).

Comment 4 Richard W.M. Jones 2012-05-17 20:43:02 UTC
Yes, confirmed that this does work with the later kernel.

Note You need to log in before you can comment on or make changes to this bug.