automount ignores remote logins

automount ignores remote logins

[Max Harvest]

[-- Attachment #1.1: Type: text/plain, Size: 2000 bytes --]

Hello everybody,

since i'm using the automount 5, i'm confronted with a very strange
behaviour. I use it to mount user directories over nfs. The automounter
maps are provided by NIS und users ususally have to login only into
client computers remotely using rsh or ssh. Now i encounter the following
missbehaviour:

At the init, automounter seems to start propperly, but if someone logins
over the network into the client machine, the automounter ignores all
mount requests (i.e. there is no automounter activity at all). Even a
restart changes nothing. But if someone logins on the client machine
over a tty, everything works fine. Therefor i enabled debug mode
verbosity and this is the output from syslog:

after the startup during boot or after a restart later:

automount[1910]: Starting automounter version 5.0.2, master map auto.master
automount[1910]: using kernel protocol version 5.00
automount[1910]: lookup_nss_read_master: reading master nis auto.master
automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.master
automount[1910]: parse_init: parse(sun): init gathered global
options: (null)
automount[1910]: mount_init: mount(bind): bind_works = 1

after login per tty:

automount[1910]: master_do_mount: mounting /users
automount[1910]: lookup_nss_read_map: reading map yp auto.users
automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.users
automount[1910]: parse_init: parse(sun): init gathered global options:
(null)
automount[1910]: mount_init: mount(bind): bind_works = 1
automount[1910]: mounted indirect mount on /users with timeout 600, freq 150
seconds
automount[1910]: ghosting enabled

It seems, that a login through tty sends an wake up event to the
automount deamon, because all remote logins are ignored. This
behaviour seems also to be independent of all different automounter
and NIS configuration variants or options. I have tried a lot of different
(propperly working over tty) configurations without any effect on this
strange behaviour.

Any idea?

[-- Attachment #1.2: Type: text/html, Size: 2286 bytes --]

[-- Attachment #2: Type: text/plain, Size: 140 bytes --]

_______________________________________________
autofs mailing list
autofs@linux.kernel.org
http://linux.kernel.org/mailman/listinfo/autofs

automount ignores remote logins

[Max Harvest]

Hello everybody,

since I'm using the automount 5, I'm confronted with a very strange
behaviour. I use it to mount user directories over nfs. The automounter
maps are provided by NIS und users usually have to login only into
client computers remotely using rsh or ssh. Now i encounter the following
misbehaviour:

At the init, automounter seems to start properly, but if someone logins
over the network into the client machine, the automounter ignores all
mount requests (i.e. there is no automounter activity at all). Even a
restart changes nothing. But if someone log ins on the client machine
over a tty, everything works fine. Therefor i enabled debug mode
verbosity and this is the output from syslog:

after the start up during boot or after a restart later:

automount[1910]: Starting automounter version 5.0.2, master map auto.master
automount[1910]: using kernel protocol version 5.00
automount[1910]: lookup_nss_read_master: reading master nis auto.master
automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.master
automount[1910]: parse_init: parse(sun): init gathered global options: (null)
automount[1910]: mount_init: mount(bind): bind_works = 1

after login per tty:

automount[1910]: master_do_mount: mounting /users
automount[1910]: lookup_nss_read_map: reading map yp auto.users
automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.users
automount[1910]: parse_init: parse(sun): init gathered global options: (null)
automount[1910]: mount_init: mount(bind): bind_works = 1
automount[1910]: mounted indirect mount on /users with timeout 600,
freq 150 seconds
automount[1910]: ghosting enabled

It seems, that a login through tty sends an wake up event to the
automount deamon, because all remote logins are ignored. This
behaviour seems also to be independent of all different automounter
and NIS configuration variants or options. I have tried a lot of different
(properly working over tty) configurations without any effect on this
strange behaviour.

Any idea?

automount ignores remote logins

[Max Harvest]

Hello everybody,

since I'm using the automount 5, I'm confronted with a very strange
behaviour. I use it to mount user directories over nfs. The automounter
maps are provided by NIS und users usually have to login only into
client computers remotely using rsh or ssh. Now i encounter the following
misbehaviour:

At the init, automounter seems to start properly, but if someone logins
over the network into the client machine, the automounter ignores all
mount requests (i.e. there is no automounter activity at all). Even a
restart changes nothing. But if someone log ins on the client machine
over a tty, everything works fine. Therefor i enabled debug mode
verbosity and this is the output from syslog:

after the start up during boot or after a restart later:

automount[1910]: Starting automounter version 5.0.2, master map auto.master
automount[1910]: using kernel protocol version 5.00
automount[1910]: lookup_nss_read_master: reading master nis auto.master
automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.master
automount[1910]: parse_init: parse(sun): init gathered global options: (null)
automount[1910]: mount_init: mount(bind): bind_works = 1

after login per tty:

automount[1910]: master_do_mount: mounting /users
automount[1910]: lookup_nss_read_map: reading map yp auto.users
automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.users
automount[1910]: parse_init: parse(sun): init gathered global options: (null)
automount[1910]: mount_init: mount(bind): bind_works = 1
automount[1910]: mounted indirect mount on /users with timeout 600,
freq 150 seconds
automount[1910]: ghosting enabled

It seems, that a login through tty sends an wake up event to the
automount deamon, because all remote logins are ignored. This
behaviour seems also to be independent of all different automounter
and NIS configuration variants or options. I have tried a lot of different
(properly working over tty) configurations without any effect on this
strange behaviour.

Any idea?

Re: automount ignores remote logins

[Jeff Moyer]

"Max Harvest" <rootserver94@googlemail.com> writes:

> Hello everybody,
>
> since i'm using the automount 5, i'm confronted with a very strange
> behaviour. I use it to mount user directories over nfs. The automounter

Are you saying that you did not experience this problem with autofs v4?

> maps are provided by NIS und users ususally have to login only into
> client computers remotely using rsh or ssh. Now i encounter the following
> missbehaviour:
>
> At the init, automounter seems to start propperly, but if someone logins
> over the network into the client machine, the automounter ignores all
> mount requests (i.e. there is no automounter activity at all). Even a

This is probably because there is no lookup performed (meaning it never
gets to autofs).  What errors are displayed in the ssh session when a
user logs in?  Is there anything in /var/log/secure or /var/log/messages
that may indicate a problem elsewhere?

-Jeff

Re: automount ignores remote logins

[Max Harvest]

On Wed, Jun 11, 2008 at 4:09 PM, Jeff Moyer <jmoyer@redhat.com> wrote:
> "Max Harvest" <rootserver94@googlemail.com> writes:
>
>> Hello everybody,
>>
>> since i'm using the automount 5, i'm confronted with a very strange
>> behaviour. I use it to mount user directories over nfs. The automounter
>
> Are you saying that you did not experience this problem with autofs v4?
>

yes. I have downgraded to version 4 and everything works fine, but  
this is not very satisfactory.

>> maps are provided by NIS und users ususally have to login only into
>> client computers remotely using rsh or ssh. Now i encounter the following
>> missbehaviour:
>>
>> At the init, automounter seems to start propperly, but if someone logins
>> over the network into the client machine, the automounter ignores all
>> mount requests (i.e. there is no automounter activity at all). Even a
>
> This is probably because there is no lookup performed (meaning it never
> gets to autofs).  What errors are displayed in the ssh session when a
> user logs in?  Is there anything in /var/log/secure or /var/log/messages
> that may indicate a problem elsewhere?

of course there are lookups. But if an ssh or rsh login occurs, the message
is 'no such file or directory'. There is no message from automount in syslog
(only from ssh, that confirms a login). It happens exactly as i wrote, a
login through tty lets automount read the indirect maps and from this point
automunt also reacts to lookups on remote logins. The strange thing is, that
reading the indirect map happens without any lookup, just a login on tty
(e.g. root login). I'm just curious which event cause the secoud part of the
output i posted earlier.

Max

Re: automount ignores remote logins

[Ian Kent]

On Wed, 2008-06-11 at 17:55 +0200, Max Harvest wrote:
> On Wed, Jun 11, 2008 at 4:09 PM, Jeff Moyer <jmoyer@redhat.com> wrote:
> > "Max Harvest" <rootserver94@googlemail.com> writes:
> >
> >> Hello everybody,
> >>
> >> since i'm using the automount 5, i'm confronted with a very strange
> >> behaviour. I use it to mount user directories over nfs. The automounter
> >
> > Are you saying that you did not experience this problem with autofs v4?
> >
> 
> yes. I have downgraded to version 4 and everything works fine, but  
> this is not very satisfactory.
> 
> >> maps are provided by NIS und users ususally have to login only into
> >> client computers remotely using rsh or ssh. Now i encounter the following
> >> missbehaviour:
> >>
> >> At the init, automounter seems to start propperly, but if someone logins
> >> over the network into the client machine, the automounter ignores all
> >> mount requests (i.e. there is no automounter activity at all). Even a
> >
> > This is probably because there is no lookup performed (meaning it never
> > gets to autofs).  What errors are displayed in the ssh session when a
> > user logs in?  Is there anything in /var/log/secure or /var/log/messages
> > that may indicate a problem elsewhere?
> 
> of course there are lookups. But if an ssh or rsh login occurs, the message
> is 'no such file or directory'. There is no message from automount in syslog
> (only from ssh, that confirms a login). It happens exactly as i wrote, a
> login through tty lets automount read the indirect maps and from this point
> automunt also reacts to lookups on remote logins. The strange thing is, that
> reading the indirect map happens without any lookup, just a login on tty
> (e.g. root login). I'm just curious which event cause the secoud part of the
> output i posted earlier.

This description and the information you provided is not sufficient to
work out what's going on.

Do this on a test machine:

Ensure autofs debug logging is enabled.
With autofs stopped (obviously you need to log into an account that
doesn't use automounted directories).
Check that no automount related mounts are active by
inspecting /proc/mounts.
Start autofs.
Attempt to log in from a remote machine using an account that does use
an automounted directory.
Attempt to log in locally using an account that does use an automounted
directory.
Log both sessions out.
Stop autofs.
Post resulting log here.

Ian

Re: automount ignores remote logins

[Ian Kent]

On Thu, 2008-06-12 at 11:14 +0200, Max Harvest wrote:
> Ian Kent wrote:
> 
> >
> > Do this on a test machine:
> >
> > Ensure autofs debug logging is enabled.
> done
> > With autofs stopped (obviously you need to log into an account that
> > doesn't use automounted directories).
> > Check that no automount related mounts are active by
> > inspecting /proc/mounts.
> done, no automount entries in /proc/mounts
> > Start autofs.
> this is the output:
> Jun  12 10:11:22 node061 automount[2274]: Starting automounter version  
> 5.0.2, master map auto.master
> Jun  12 10:11:22 node061 automount[2274]: using kernel protocol version 5.00
> Jun  12 10:11:22 node061 automount[2274]: lookup_nss_read_master:  
> reading master nis auto.master
> Jun  12 10:11:22 node061 automount[2274]: lookup_init: lookup(yp):  
> ctxt->mapname=auto.master
> Jun  12 10:11:22 node061 automount[2274]: parse_init: parse(sun): init  
> gathered global options: (null)
> Jun  12 10:11:22 node061 automount[2274]: mount_init: mount(bind):  
> bind_works = 1
> 
> > Attempt to log in from a remote machine using an account that does use
> > an automounted directory.
> this is the output:
> Jun  12 10:11:53 node061 sshd[2304]: Accepted keyboard-interactive/pam  
> for maxh from 192.168.0.250 port 44822 ssh2
> I can log in, but have no home. There is no output from automount.
> 
> > Attempt to log in locally using an account that does use an automounted
> > directory.
> works (as allready posted), the account name is "maxh" this is the output:
> Jun  12 10:17:07 node061 automount[2274]: master_do_mount: mounting /users
> Jun  12 10:17:07 node061 automount[2274]: lookup_nss_read_map: reading  
> map yp auto.users
> Jun  12 10:17:07 node061 automount[2274]: lookup_init: lookup(yp):  
> ctxt->mapname=auto.users
> Jun  12 10:17:07 node061 automount[2274]: parse_init: parse(sun): init  
> gathered global options: (null)
> Jun  12 10:17:07 node061 automount[2274]: mount_init: mount(bind):  
> bind_works = 1
> Jun  12 10:17:07 node061 automount[2274]: mounted indirect mount on  
> /users with timeout 600, freq 150 seconds
> Jun  12 10:17:07 node061 automount[2274]: ghosting enabled
> Jun  12 10:17:12 node061 automount[2274]: handle_packet: type = 3
> Jun  12 10:17:12 node061 automount[2274]:  
> handle_packet_missing_indirect: token 1, name maxh, request pid 2325
> Jun  12 10:17:12 node061 automount[2274]: attempting to mount entry  
> /users/maxh
> Jun  12 10:17:12 node061 automount[2274]: lookup_mount: lookup(file):  
> looking up maxh
> Jun  12 10:17:12 node061 automount[2274]: lookup_mount: lookup(file):  
> maxh -> bootserver:/home/maxh
> Jun  12 10:17:12 node061 automount[2274]: parse_mount: parse(sun):  
> expanded entry: bootserver:/home/maxh
> Jun  12 10:17:12 node061 automount[2274]: parse_mount: parse(sun):  
> gathered options: fstype=nfs,soft,intr
> Jun  12 10:17:12 node061 automount[2274]: parse_mount: parse(sun):  
> dequote("bootserver:/home/maxh") -> bootserver:/home/maxh
> Jun  12 10:17:12 node061 automount[2274]: parse_mount: parse(sun):  
> core of entry: options=fstype=nfs,soft,intr, loc=bootserver:/home/maxh
> Jun  12 10:17:12 node061 automount[2274]: sun_mount: parse(sun):  
> mounting root /users, mountpoint maxh, what bootserver:/home/maxh,  
> fstype nfs, options soft,intr
> Jun  12 10:17:12 node061 automount[2274]: mount_mount: mount(nfs):  
> root=/users name=maxh what=bootserver:/home/maxh, fstype=nfs,  
> options=soft,intr
> Jun  12 10:17:12 node061 automount[2274]: mount_mount: mount(nfs): nfs  
> options="soft,intr", nosymlink=0, ro=0
> Jun  12 10:17:12 node061 automount[2274]: mount_mount: mount(nfs):  
> calling mkdir_path /users/maxh
> Jun  12 10:17:12 node061 automount[2274]: mount_mount: mount(nfs):  
> calling mount -t nfs -s -o soft,intr bootserver:/home/maxh /users/maxh
> Jun  12 10:17:12 node061 automount[2274]: Version 1.1.0 Starting
> Jun  12 10:17:12 node061 automount[2274]: Flags:
> Jun  12 10:17:12 node061 automount[2274]: statd running as root. chown  
> /var/lib/nfs/sm to choose different user
> Jun  12 10:17:12 node061 automount[2274]: mount(nfs): mounted  
> bootserver:/home/maxh on /users/maxh
> Jun  12 10:17:12 node061 automount[2274]: send_ready: token = 1
> Jun  12 10:17:12 node061 automount[2274]: mounted /users/maxh
> 
> > Log both sessions out.
> done. The output from remote session:
> Jun  12 10:19:49 node061 sshd[2348]: PAM audit_log_acct_message()  
> failed: Operation not permitted
> 
> > Stop autofs.
> works. This is the output:
> Jun  12 10:22:52 node061 automount[2274]: do_notify_state: signal 15
> Jun  12 10:22:52 node061 automount[2274]: master_notify_state_change:  
> sig 15 switching /users from 1 to 5
> Jun  12 10:22:52 node061 automount[2274]: st_prepare_shutdown: state 1  
> path /users
> Jun  12 10:22:52 node061 automount[2274]: expire_proc: exp_proc =  
> 1075984720 path /users
> Jun  12 10:22:52 node061 automount[2274]: expire_proc_indirect: expire  
> /users/maxh
> Jun  12 10:22:52 node061 automount[2274]: handle_packet: type = 4
> Jun  12 10:22:52 node061 automount[2274]:  
> handle_packet_expire_indirect: token 2, name maxh
> Jun  12 10:22:52 node061 automount[2274]: expiring path /users/maxh
> Jun  12 10:22:52 node061 automount[2274]: umount_multi: path  
> /users/maxh incl 1
> Jun  12 10:22:52 node061 automount[2274]: unmounting dir = /users/maxh
> Jun  12 10:22:52 node061 automount[2274]: rm_unwanted_fn: removing  
> directory /users/maxh
> Jun  12 10:22:52 node061 automount[2274]: expired /users/maxh
> Jun  12 10:22:52 node061 automount[2274]: send_ready: token = 2
> Jun  12 10:22:53 node061 automount[2274]: expire_cleanup: got thid  
> 1075984720 path /users stat 0
> Jun  12 10:22:53 node061 automount[2274]: expire_cleanup: sigchld: exp  
> 1075984720 finished, switching from 5 to 7
> Jun  12 10:22:53 node061 automount[2274]: umount_multi: path /users incl 0
> Jun  12 10:22:53 node061 automount[2274]: rm_unwanted_fn: removing  
> directory /users/work
> Jun  12 10:22:53 node061 automount[2274]: umounted indirect mount /users
> Jun  12 10:22:53 node061 automount[2274]: shut down path /users
> 
> > Post resulting log here.
> see above.

No, this is not what I asked for.
Go through the procedure and post the log as a single unit.
Add comments about the activities that relate to timestamps in the log
but I want to see a single continuous log. If you have to post the log
as an attachment.

Ian

Re: automount ignores remote logins

[Ian Kent]

On Thu, 2008-06-12 at 15:45 +0200, Max Harvest wrote:
> Quoting Ian Kent <raven@themaw.net>:
> > No, this is not what I asked for.
> > Go through the procedure and post the log as a single unit.
> > Add comments about the activities that relate to timestamps in the log
> > but I want to see a single continuous log. If you have to post the log
> > as an attachment.
> 
> The complete /var/log/messages (from boot to shutdown) is attached.  
> Comments are indicated by #

# root starts automount
Jun 12 15:13:02 node061 automount[1910]: Starting automounter version 5.0.2, master map auto.master
Jun 12 15:13:02 node061 automount[1910]: using kernel protocol version 5.00
Jun 12 15:13:02 node061 automount[1910]: lookup_nss_read_master: reading master nis auto.master
Jun 12 15:13:02 node061 automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.master
Jun 12 15:13:02 node061 automount[1910]: parse_init: parse(sun): init gathered global options: (null)
Jun 12 15:13:02 node061 automount[1910]: mount_init: mount(bind): bind_works = 1

But automount has not read the nis master map or it is empty or it
hasn't completed it's startup. It hasn't mounted any master map
automounts.

# root performs local log in (tty)
Jun 12 15:17:07 node061 automount[1910]: master_do_mount: mounting /users
Jun 12 15:17:07 node061 automount[1910]: lookup_nss_read_map: reading map yp auto.users
Jun 12 15:17:07 node061 automount[1910]: lookup_init: lookup(yp): ctxt->mapname=auto.users
Jun 12 15:17:07 node061 automount[1910]: parse_init: parse(sun): init gathered global options: (null)
Jun 12 15:17:07 node061 automount[1910]: mount_init: mount(bind): bind_works = 1
Jun 12 15:17:07 node061 automount[1910]: mounted indirect mount on /users with timeout 600, freq 150 seconds
Jun 12 15:17:07 node061 automount[1910]: ghosting enabled

Four minutes pass before this log entry, which appears to be the
remainder of the startup, assuming we are seeing everything in the log
for this four minute period, with only the following entry and
apparently nothing else:

# user maxh ($HOME=/users/maxh) performs remote log in _and_ log out
Jun 12 15:13:53 node061 sshd[1999]: Accepted keyboard-interactive/pam for maxh from 192.168.0.250 port 44822 ssh2
Jun 12 15:14:55 node061 sshd[2002]: PAM audit_log_acct_message() failed: Operation not permitted

For which there is no automount activity presumably because there are no
autofs mounts mounted at this time.

You are the person on site, so you are best placed to understand what is
happening.

I have no idea why autofs is either not seeing any nis master map
entries or is blocking on network io at startup. However, we don't see
evidence of a HUP signal either so the log entries starting at:
Jun 12 15:17:07 node061 automount[1910]: master_do_mount: mounting /users
are probably the tail end of the startup rather than the result of a
signal to re-read the map.

The gaps in time stamps in the log make me suspicious that we're not
getting the whole story. Is that really the case?

Ian

Re: automount ignores remote logins

[Ian Kent]

On Thu, 2008-06-12 at 16:59 +0200, Max Harvest wrote:
> > The gaps in time stamps in the log make me suspicious that we're not
> > getting the whole story. Is that really the case?
> The gaps result from (slow) interaction. The computer is a diskless  
> client and the local and remote console are in different locations.  
> The log is complete.

If it is diskless, what is the root fs.

Ian