Skip to contentSkip to conversation

Unmount not working with 2.6.21 & 4.1.3/4.1.4

DISCUSSION Jarl Stefansson 3 participants 16 messages Mar 25, 2007 – Apr 6, 2007 1 message not in the archive Kernel automounter

lore
Share & tools
lore
#1 lore raw
In reply to a message not in the archive

I'm working on a custom Gentoo build and autofs seems to completely ignore all timeout/unmount attempts.

My configuration is 2.6.21-rc3-mm2 with 4.1.3-R7.

The mounting works well but devices never get unmounted, even sending SIGUSR1 or killing does not have any effect. There is no relevant logging to /var/log/messages even when running with debug enabled.

My Config:

/usr/sbin/automount -dgt 3 /mnt/auto file /etc/autofs/auto.mount

ubadata -fstype=ext2,ro,async,exec,noatime,nodev :/dev/uba1

I also included /var/log/messages from starting automount and mounting a device if that can be of any help.

Jarl

######################################### /var/log/messages ####################################################

Mar 26 01:33:43 kidsq daemon.info automount[3608]: starting automounter version 4.1.3, path = /mnt/auto, maptype = file, mapname = /etc/autofs/auto.mount
Mar 26 01:33:43 kidsq daemon.debug automount[3608]: mount(bind): bind_works = 1 
Mar 26 01:33:43 kidsq daemon.info automount[3608]: using kernel protocol version 4.00
Mar 26 01:33:43 kidsq daemon.info automount[3608]: using timeout 3 seconds; freq 1 secs
Mar 26 01:33:43 kidsq daemon.info automount[3608]: ghosting enabled
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3612
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: handle_child: got pid 3612, sig 0 (0), stat 0 
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: sigchld: exp 3612 finished, switching from 2 to 1
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:44 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3613
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: handle_child: got pid 3613, sig 0 (0), stat 0 
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: sigchld: exp 3613 finished, switching from 2 to 1
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:45 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3614
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: handle_child: got pid 3614, sig 0 (0), stat 0 
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: sigchld: exp 3614 finished, switching from 2 to 1
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:46 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3615
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: handle_child: got pid 3615, sig 0 (0), stat 0 
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: sigchld: exp 3615 finished, switching from 2 to 1
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:47 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3616
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: handle_child: got pid 3616, sig 0 (0), stat 0 
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: sigchld: exp 3616 finished, switching from 2 to 1
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:48 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3617
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: handle_child: got pid 3617, sig 0 (0), stat 0 
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: sigchld: exp 3617 finished, switching from 2 to 1
Mar 26 01:33:49 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:50 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3618
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: handle_child: got pid 3618, sig 0 (0), stat 0 
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: sigchld: exp 3618 finished, switching from 2 to 1
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:51 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3619
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: handle_child: got pid 3619, sig 0 (0), stat 0 
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: sigchld: exp 3619 finished, switching from 2 to 1
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:52 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3620
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: handle_child: got pid 3620, sig 0 (0), stat 0 
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: sigchld: exp 3620 finished, switching from 2 to 1
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: handle_packet: type = 0 
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: handle_packet_missing: token 46, name ubadata 
Mar 26 01:33:53 kidsq daemon.info automount[3608]: attempting to mount entry /mnt/auto/ubadata
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: lookup(file): ubadata -> -fstype=auto,rw,sync,noexec,nodev^I:/dev/uba1
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: parse(sun): expanded entry: -fstype=auto,rw,sync,noexec,nodev^I:/dev/uba1
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: parse(sun): dequote("fstype=auto,rw,sync,noexec,nodev") -> fstype=auto,rw,sync,noexec,nodev
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: parse(sun): gathered options: fstype=auto,rw,sync,noexec,nodev
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: parse(sun): dequote("/dev/uba1") -> /dev/uba1
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: parse(sun): core of entry: options=fstype=auto,rw,sync,noexec,nodev, loc=/dev/uba1
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: parse(sun): mounting root /mnt/auto, mountpoint ubadata, what /dev/uba1, fstype auto, options rw,sync,noexec,nodev 
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: do_mount /dev/uba1 /mnt/auto/ubadata type auto options rw,sync,noexec,nodev using module generic
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: mount(generic): calling mkdir_path /mnt/auto/ubadata
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: mount(generic): calling mount -t auto -o rw,sync,noexec,nodev /dev/uba1 /mnt/auto/ubadata
Mar 26 01:33:53 kidsq user.info kernel: kjournald starting.  Commit interval 5 seconds
Mar 26 01:33:53 kidsq user.info kernel: EXT3 FS on uba1, internal journal
Mar 26 01:33:53 kidsq user.info kernel: EXT3-fs: mounted filesystem with ordered data mode.
Mar 26 01:33:53 kidsq daemon.debug automount[3622]: mount(generic): mounted /dev/uba1 type auto on /mnt/auto/ubadata
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: handle_child: got pid 3622, sig 0 (0), stat 0 
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: sig_child: found pending iop pid 3622: signalled 0 (sig 0), exit status 0
Mar 26 01:33:53 kidsq daemon.debug automount[3608]: send_ready: token=46 
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3626
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: handle_child: got pid 3626, sig 0 (0), stat 1 
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: sigchld: exp 3626 finished, switching from 2 to 1
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:54 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:55 kidsq daemon.debug automount[3627]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3627
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: handle_child: got pid 3627, sig 0 (0), stat 1 
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: sigchld: exp 3627 finished, switching from 2 to 1
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:55 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:56 kidsq daemon.debug automount[3628]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3628
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: handle_child: got pid 3628, sig 0 (0), stat 1 
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: sigchld: exp 3628 finished, switching from 2 to 1
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:56 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:57 kidsq daemon.debug automount[3629]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3629
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: handle_child: got pid 3629, sig 0 (0), stat 1 
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: sigchld: exp 3629 finished, switching from 2 to 1
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:57 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:58 kidsq daemon.debug automount[3630]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3630
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: handle_child: got pid 3630, sig 0 (0), stat 1 
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: sigchld: exp 3630 finished, switching from 2 to 1
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:58 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:33:59 kidsq daemon.debug automount[3631]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3631
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: handle_child: got pid 3631, sig 0 (0), stat 1 
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: sigchld: exp 3631 finished, switching from 2 to 1
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:33:59 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:34:00 kidsq daemon.debug automount[3632]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3632
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: handle_child: got pid 3632, sig 0 (0), stat 1 
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: sigchld: exp 3632 finished, switching from 2 to 1
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:34:00 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:34:01 kidsq daemon.debug automount[3633]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3633
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: handle_child: got pid 3633, sig 0 (0), stat 1 
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: sigchld: exp 3633 finished, switching from 2 to 1
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:34:01 kidsq daemon.debug automount[3608]: st_ready(): state = 2 
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: sig 14 switching from 1 to 2
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: get_pkt: state 1, next 2
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: st_expire(): state = 1 
Mar 26 01:34:02 kidsq daemon.debug automount[3634]: expire_proc: 1 remaining in /mnt/auto 
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: expire_proc: exp_proc=3634
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: handle_child: got pid 3634, sig 0 (0), stat 1 
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: sigchld: exp 3634 finished, switching from 2 to 1
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: get_pkt: state 2, next 1
Mar 26 01:34:02 kidsq daemon.debug automount[3608]: st_ready(): state = 2 

(this sequence keeps going forever)

maintainer
#2 lore raw
Quoted 3 lines · Jarl Stefansson

I'm working on a custom Gentoo build and autofs seems to completely ignore all timeout/unmount attempts.

My configuration is 2.6.21-rc3-mm2 with 4.1.3-R7.

Are you sure this doesn't happen with 2.6.20?

Ian

maintainer
#3 lore raw
Quoted 6 lines · Ian Kent
Quoted 3 lines · Jarl Stefansson

I'm working on a custom Gentoo build and autofs seems to completely ignore all timeout/unmount attempts.

My configuration is 2.6.21-rc3-mm2 with 4.1.3-R7.

Are you sure this doesn't happen with 2.6.20?

And does this also happen with 2.6.21-rc kernels?

Ian

#4 lore raw

Hi Ian,

I spent the morning testing automount on several other kernels, I have verified that the problem also affects:

2.6.19-gentoo-r1
2.6.20-gentoo-r5
2.6.21-rc2-git4

Yours

Jarl

----- Original Message -----
From: Ian Kent <[email protected]>
To: Jarl Stefansson <[email protected]>
Cc: [email protected]
Sent: Monday, March 26, 2007 5:13:09 AM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Quoted 6 lines · Ian Kent
Quoted 3 lines · Jarl Stefansson

I'm working on a custom Gentoo build and autofs seems to completely ignore all timeout/unmount attempts.

My configuration is 2.6.21-rc3-mm2 with 4.1.3-R7.

Are you sure this doesn't happen with 2.6.20?

And does this also happen with 2.6.21-rc kernels?

Ian

maintainer
#6 lore raw
In reply to Jarl Stefansson
Quoted 6 lines · Jarl Stefansson

I'm working on a custom Gentoo build and autofs seems to completely ignore all timeout/unmount attempts.

My configuration is 2.6.21-rc3-mm2 with 4.1.3-R7.

The mounting works well but devices never get unmounted, even sending SIGUSR1 or killing does not have any effect. There is no relevant logging to /var/log/messages even when running with debug enabled.

On second thoughts, before we go down the kernel path, maybe I should ask a bit more about the history.

Have you had autofs up and running OK on this setup before?

Ian

maintainer
#5 lore raw
In reply to Jarl Stefansson
Quoted 7 lines · Jarl Stefansson

Hi Ian,

I spent the morning testing automount on several other kernels, I have verified that the problem also affects:

2.6.19-gentoo-r1
2.6.20-gentoo-r5
2.6.21-rc2-git4

OK good, then it's not likely the changes that went into 2.6.21.

So we need an answer to my other question to move forward.

Ian

#7 lore raw
In reply to Ian Kent

Hi Ian,

This is a brand new box so no.

The automount app is standard gentoo 4.1.3-R7, I also tried 4.1.4 but I had issues getting it to run (with the same config as 4.1.3 it simply would not start)

I googled for two days and verified that all known issues/race conditions seem to be patched in 4.1.3-R7 so I'm at a loss as to what the problem might be.

Yours

Jarl

----- Original Message -----
From: Ian Kent <[email protected]>
To: Jarl Stefansson <[email protected]>
Cc: [email protected]
Sent: Monday, March 26, 2007 2:16:24 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Quoted 6 lines · Jarl Stefansson

I'm working on a custom Gentoo build and autofs seems to completely ignore all timeout/unmount attempts.

My configuration is 2.6.21-rc3-mm2 with 4.1.3-R7.

The mounting works well but devices never get unmounted, even sending SIGUSR1 or killing does not have any effect. There is no relevant logging to /var/log/messages even when running with debug enabled.

On second thoughts, before we go down the kernel path, maybe I should ask a bit more about the history.

Have you had autofs up and running OK on this setup before?

Ian

#8 lore raw

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hi Ian, This is a brand new box so no.

jarl> The automount app is standard gentoo 4.1.3-R7, I also tried 4.1.4 but jarl> I had issues getting it to run (with the same config as 4.1.3 it jarl> simply would not start)

jarl> I googled for two days and verified that all known issues/race jarl> conditions seem to be patched in 4.1.3-R7 so I'm at a loss as to what jarl> the problem might be.

Are you sure that some application doesn't have an active reference to the automounted file system?

-Jeff

#9 lore raw

Hey Jeff,

Yes, I am 100% sure that there is nothing using the mounted filesystems. For debugging I have a system with just bare minimals, kernel + glibc + base utilities so there really aren't any applications which can interfere.

According to what I have read I should also be getting some debug info in the log regarding the timeout period but I can't see any mention of timeout or unmount.

Yours

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 4:58:17 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hi Ian, This is a brand new box so no.

jarl> The automount app is standard gentoo 4.1.3-R7, I also tried 4.1.4 but jarl> I had issues getting it to run (with the same config as 4.1.3 it jarl> simply would not start)

jarl> I googled for two days and verified that all known issues/race jarl> conditions seem to be patched in 4.1.3-R7 so I'm at a loss as to what jarl> the problem might be.

Are you sure that some application doesn't have an active reference to the automounted file system?

-Jeff

#10 lore raw

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

#11 lore raw

Hey Jeff,

Here is the debug log when set to 60 sec

http://pastebin.ca/411870

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 7:29:18 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

maintainer
#12 lore raw
Quoted 5 lines · Jarl Stefansson

Hey Jeff,

Here is the debug log when set to 60 sec

http://pastebin.ca/411870

Unfortunately, that hasn't helped.
Version 4 doesn't log the pid of the process accessing the mount point, so we still don't know what's causing this.

Do you have hald running?
Does this behavior change after an hour or so?

Are there many processes running?
If not perhaps a list of them would help.

Quoted 31 lines

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 7:29:18 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

#13 lore raw

Hey Ian,

Here is a list of the processes running:

http://pastebin.ca/412160

Jarl

----- Original Message -----
From: Ian Kent <[email protected]>
To: Jarl Stefansson <[email protected]>
Cc: [email protected], [email protected]
Sent: Tuesday, March 27, 2007 2:52:21 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Quoted 5 lines · Jarl Stefansson

Hey Jeff,

Here is the debug log when set to 60 sec

http://pastebin.ca/411870

Unfortunately, that hasn't helped.
Version 4 doesn't log the pid of the process accessing the mount point, so we still don't know what's causing this.

Do you have hald running?
Does this behavior change after an hour or so?

Are there many processes running?
If not perhaps a list of them would help.

Quoted 31 lines

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 7:29:18 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

#14 lore raw

I'm still at a loss as to why automounted file-systems are not being unmounted.

Any suggestions on how to proceed with debugging this issue are appreciated, if there is any additional information I can provide please don't hesitate to ask.

Yours
Jarl

----- Original Message -----
From: Jarl Stefansson <[email protected]>
To: Ian Kent <[email protected]>
Cc: [email protected]
Sent: Tuesday, March 27, 2007 3:10:52 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Hey Ian,

Here is a list of the processes running:

http://pastebin.ca/412160

Jarl

----- Original Message -----
From: Ian Kent <[email protected]>
To: Jarl Stefansson <[email protected]>
Cc: [email protected], [email protected]
Sent: Tuesday, March 27, 2007 2:52:21 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Quoted 5 lines · Jarl Stefansson

Hey Jeff,

Here is the debug log when set to 60 sec

http://pastebin.ca/411870

Unfortunately, that hasn't helped.
Version 4 doesn't log the pid of the process accessing the mount point, so we still don't know what's causing this.

Do you have hald running?
Does this behavior change after an hour or so?

Are there many processes running?
If not perhaps a list of them would help.

Quoted 23 lines

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 7:29:18 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 
Quoted 8 lines

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs

maintainer
#15 lore raw
Quoted 4 lines · Jarl Stefansson

I'm still at a loss as to why automounted file-systems are not being unmounted.

Any suggestions on how to proceed with debugging this issue are appreciated, if there is any additional information I can provide please don't hesitate to ask.

I don't have any ideas about how to find out what is accessing your mount.

lsof may give some information but it looks like the frequency of access is high and probably brief (or there's an open file or working directory within the automount), as there is no umount followed by mount in the log.

Perhaps /mnt is a bad place to put this automount in Gentoo. Try another location.

Obviously, this isn't a usb device because as you know, once they're mounted, they can often be polled frequently. Don't ask me what software or why, I don't know.

Quoted 85 lines

Yours
Jarl

----- Original Message -----
From: Jarl Stefansson <[email protected]>
To: Ian Kent <[email protected]>
Cc: [email protected]
Sent: Tuesday, March 27, 2007 3:10:52 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Hey Ian,

Here is a list of the processes running:

http://pastebin.ca/412160

Jarl

----- Original Message -----
From: Ian Kent <[email protected]>
To: Jarl Stefansson <[email protected]>
Cc: [email protected], [email protected]
Sent: Tuesday, March 27, 2007 2:52:21 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Quoted 5 lines · Jarl Stefansson

Hey Jeff,

Here is the debug log when set to 60 sec

http://pastebin.ca/411870

Unfortunately, that hasn't helped.
Version 4 doesn't log the pid of the process accessing the mount point, so we still don't know what's causing this.

Do you have hald running?
Does this behavior change after an hour or so?

Are there many processes running?
If not perhaps a list of them would help.

Quoted 23 lines

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 7:29:18 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 
Quoted 8 lines

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs

_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs

maintainer
#16 lore raw
In reply to Jarl Stefansson
Quoted 4 lines · Jarl Stefansson

I'm still at a loss as to why automounted file-systems are not being unmounted.

Any suggestions on how to proceed with debugging this issue are appreciated, if there is any additional information I can provide please don't hesitate to ask.

I just installed a basic Gentoo 2006.1 with gentoo-source kernel r5 and autofs-4.1.3-r7. I setup a simple bind mount config and set timeout to 3 seconds.

This functions as expected.

Quoted 86 lines

Yours
Jarl

----- Original Message -----
From: Jarl Stefansson <[email protected]>
To: Ian Kent <[email protected]>
Cc: [email protected]
Sent: Tuesday, March 27, 2007 3:10:52 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Hey Ian,

Here is a list of the processes running:

http://pastebin.ca/412160

Jarl

----- Original Message -----
From: Ian Kent <[email protected]>
To: Jarl Stefansson <[email protected]>
Cc: [email protected], [email protected]
Sent: Tuesday, March 27, 2007 2:52:21 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

Quoted 5 lines · Jarl Stefansson

Hey Jeff,

Here is the debug log when set to 60 sec

http://pastebin.ca/411870

Unfortunately, that hasn't helped.
Version 4 doesn't log the pid of the process accessing the mount point, so we still don't know what's causing this.

Do you have hald running?
Does this behavior change after an hour or so?

Are there many processes running?
If not perhaps a list of them would help.

Quoted 23 lines

Jarl

----- Original Message -----
From: [email protected]
To: Jarl Stefansson <[email protected]>
Cc: Ian Kent <[email protected]>, [email protected]
Sent: Monday, March 26, 2007 7:29:18 PM GMT Africa/Casablanca
Subject: Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4

==> Regarding Re: [autofs] Unmount not working with 2.6.21 & 4.1.3/4.1.4; Jarl Stefansson <[email protected]> adds:

jarl> Hey Jeff, Yes, I am 100% sure that there is nothing using the mounted jarl> filesystems. For debugging I have a system with just bare minimals, jarl> kernel + glibc + base utilities so there really aren't any jarl> applications which can interfere.

jarl> According to what I have read I should also be getting some debug jarl> info in the log regarding the timeout period but I can't see any jarl> mention of timeout or unmount.

  Mar 26 01:33:54 kidsq daemon.debug automount[3626]: expire_proc: 1 remaining in /mnt/auto 
Quoted 8 lines

This means that (in your case) the kernel thought that the mount point was in use. Could you humor me and up the timeout to something like 60 seconds and see if that helps matters?

Thanks,

Jeff

_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs

_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs