portmap does not stop during shutdown, causing possible root fs corruption

On my machine portmap is started/running anyhow (together with rpc.statd), also if no nfs share has been mounted ever. In Lucid I have just installed 'nfs-common' (not the kernel server!). It is only used when I temporarly mount some nfs3-exports from guests in VBox, or nfs4-exports from my nas (running Lenny-armel).

You need to install nfs-common also if you only want just the client - and that brings portmap and rpc.statd.
Amaizingly; stopping portmap also stops rpc.statd?

Portmap is not included in nfs-common, it's a separate package. But it is refereced by nfs-common as "required"!

Setting this to Triaged. The fix is clear: we need to stop portmap after umountnfs runs.

Adding sysvinit as it currently provides no facility to stop things right after umountnfs, which portmap's old init.d script clearly defined as the point at which it should be stopped.

Unfortunately sysvinit's package import is still broken.

Attached is the debdiff to add emitting 'unmounted-remote-filesystems' to umountnfs.sh

Ok the merge proposal submitted fixes the portmap upstart job, and the debdiff will provide the event. Neither malfunctions without the other in any way, so its not critical that they are fixed at the same time, but it won't be possible to test the fix without both changes.

This bug was fixed in the package portmap - 6.0.0-2ubuntu4

---------------
portmap (6.0.0-2ubuntu4) natty; urgency=low

  [ Steve Langasek ]
  * rename debian/portmap.portmap.upstart to debian/portmap.upstart; this
    will let debhelper DTRT with both the upstart job and the init script
    under the new policy.

  [ Clint Byrum ]
  * debian/portmap.upstart: stop on unmounted-remote-filesystems, a new
    event emitted by umountnfs.sh. (LP: #711425)
 -- Dave Walker (Daviey) <email address hidden> Wed, 02 Mar 2011 13:51:55 +0000

The debdiff was sponsored - I am unsubscribing ubuntu-sponsors.

subscribing ubuntu-sponsors again - portmap was sponsored, but not sysvinit.

@Clint, your debdiff for sysvinit FTBFS:

gcc -ansi -W -Wall -O2 -fomit-frame-pointer -D_GNU_SOURCE -c -o bootlogd.o bootlogd.c
bootlogd.c: In function 'consolename':
bootlogd.c:264:15: error: 'ENOIOCTLCMD' undeclared (first use in this function)
bootlogd.c:264:15: note: each undeclared identifier is reported only once for each function it appears in
bootlogd.c: In function 'findtty':
bootlogd.c:116:8: warning: ignoring return value of 'chdir', declared with attribute warn_unused_result
bootlogd.c:131:10: warning: ignoring return value of 'chdir', declared with attribute warn_unused_result
bootlogd.c:142:10: warning: ignoring return value of 'chdir', declared with attribute warn_unused_result
bootlogd.c:147:10: warning: ignoring return value of 'chdir', declared with attribute warn_unused_result
bootlogd.c:154:7: warning: ignoring return value of 'chdir', declared with attribute warn_unused_result
make[1]: *** [bootlogd.o] Error 1
make[1]: Leaving directory `/tmp/buildd/sysvinit-2.87dsf/src'
make: *** [build-stamp] Error 2
dpkg-buildpackage: error: debian/rules build gave error exit status 2

Updated debdiff, fixes FTBFS for Natty. Wanting ack from foundations team before uploading.

Thanks.

This bug was fixed in the package sysvinit - 2.87dsf-4ubuntu21

---------------
sysvinit (2.87dsf-4ubuntu21) natty; urgency=low

  * debian/initscripts/etc/init.d/umountnfs.sh: emit a new event,
    unmounted-remote-filesystems, to allow stopping portmap
    and others. (LP: #711425)
 -- Clint Byrum <email address hidden> Fri, 04 Mar 2011 12:18:15 +0000

This is just to avoid expiry of this bug, affecting all that use nfs.

Excerpts from ingo's message of Fri May 20 17:41:49 UTC 2011:
> This is just to avoid expiry of this bug, affecting all that use nfs.
>

Ingo, only "Incomplete" tasks expire. None of the tasks associated with
this bug are Incomplete.

I did just notice that the milestone was not set for these. We definitely
want to fix this before 10.04.3 is cut and released in ~10 weeks. Setting
milestone accordingly.

With other words:

*REST in PEACE*

on this huge graveyard named Launchpad

Excerpts from ingo's message of Tue Jul 19 14:04:10 UTC 2011:
> With other words:
>
> *REST in PEACE*
>
> on this huge graveyard named Launchpad
>

Ingo, thanks for staying with us through the long spooky night.

There are some ghosts that need vanquishing before we can move on
this one.

Namely bug #665185 needs to be verified before we add this bug on top
of it.

As bug 665185 is now fixed in lucid, I guess this is no longer blocked?

Hello Clint, or anyone else affected,

Accepted sysvinit into lucid-proposed, the package will build now and be available in a few hours. Please test and give feedback here. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you in advance!

Hello Clint, or anyone else affected,

Accepted portmap into lucid-proposed, the package will build now and be available in a few hours. Please test and give feedback here. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you in advance!

Maverick survived most of its live without this fix, and we hardly get any feedback for it any more, so wontfixing.

Hello Clint,

great to hear that this bug has resurrected for beeing cured!

I did test the 'portmap' package from lucid-proposed. It also pulls in package 'initscripts' and all is fine now (before I regularly obtained 4 orphaned inodes).

Test was done in a VM running Lucid-amd64 with 2 CPU's (I do not have it on real hardware anymore).

So from my point of view it's fixed finally.

This bug was fixed in the package portmap - 6.0.0-1ubuntu2.2

---------------
portmap (6.0.0-1ubuntu2.2) lucid-proposed; urgency=low

  * debian/portmap.upstart: stop on unmounted-remote-filesystems, a new
    event emitted by umountnfs.sh. (LP: #711425)
  * d/control: Use versioned dep on initscripts to ensure portmap is
    stopped at the appropriate moment.
 -- Clint Byrum <email address hidden> Fri, 20 Jan 2012 01:37:05 -0800

This bug was fixed in the package sysvinit - 2.87dsf-4ubuntu17.5

---------------
sysvinit (2.87dsf-4ubuntu17.5) lucid-proposed; urgency=low

  * debian/initscripts/etc/init.d/umountnfs.sh: emit a new event,
    unmounted-remote-filesystems, to allow stopping portmap
    and others. (LP: #711425)
 -- Clint Byrum <email address hidden> Fri, 20 Jan 2012 01:15:42 -0800