Aitofs: spurious (?) jog messages and zombies from automountd

[Hans Ottevanger]

Hi,

When accessing a directory for the first time on /net, automounted using 
the new autofs, I reproducibly get error messages in the log.

I have exported a single directory (read/write) from my test server:

$ showmount -e soekris

Exports list on soekris:
/home

My client runs 10-STABLE, also known as 10.1-BETA2 these days, updated 
to r272021. The kernel config is GENERIC minus devices that I do not 
have, with AUTOFS added.

When accessing the server for the first time using:

ls /net/soekris/home/hans

I get the expected directory listing but immediately also see the 
following in /var/log/messages:

Sep 24 07:39:09 beastie automountd[1554]: child process 2096 terminated 
with exit status 1
Sep 24 07:39:09 beastie automountd[1554]: child process 2101 terminated 
with exit status 1

Additionally, a zombie process keep hanging forever:

$ pstree 1554
-+= 01554 root /usr/sbin/automountd
  \--- 02106 root <defunct>

When I wait until the directory is autounmounted and retry the access, 
the zombie disappears (apparently finally waited for) and I get two 
others in return:

$ pstree 1554
-+= 01554 root /usr/sbin/automountd
  |--- 02658 root <defunct>
  \--- 02663 root <defunct>

and a few extra log messages:

Sep 24 08:52:17 beastie automountd[1554]: child process 2106 terminated 
with exit status 1
Sep 24 08:52:17 beastie automountd[1554]: child process 2653 terminated 
with exit status 1

Is anyone else seeing this?
Do I miss something?
Is this expected behavior?

Kind regards,

Hans