Failed to create pty - disabling logging for job
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
upstart (Ubuntu) |
New
|
Undecided
|
Unassigned |
Bug Description
I'm booting a freshly bootstraped Precise with nfs-root. For that to work I had to replace mountall so the system isn't 100% pristine. I'm also using a custom kernel and no initramfs. But I've added some debug outputs to /dev/console there to see if everything works:
[ 0.000000] Linux version 2.6.32.
[ 0.000000] Command line: root=/dev/nfs rw nfsroot=
...
[ 38.048148] Freeing unused kernel memory: 528k freed
### after my own mount-virtfs job instead mountall:
### ls -lh /dev >/dev/console
total 0
crw------- 1 root root 5, 1 May 30 18:16 console
lrwxrwxrwx 1 root root 11 May 30 18:16 core -> /proc/kcore
lrwxrwxrwx 1 root root 13 May 30 18:16 fd -> /proc/self/fd
brw------- 1 root root 7, 0 May 30 18:16 loop0
drwxr-xr-x 2 root root 40 May 30 18:16 net
crw-rw-rw- 1 root root 1, 3 May 30 18:16 null
crw------- 1 root root 108, 0 May 30 18:16 ppp
crw-rw-rw- 1 root tty 5, 2 May 30 18:16 ptmx
drwxr-xr-x 2 root root 0 May 30 18:16 pts
drwxrwxrwt 2 root root 40 May 30 18:16 shm
lrwxrwxrwx 1 root root 15 May 30 18:16 stderr -> /proc/self/fd/2
lrwxrwxrwx 1 root root 15 May 30 18:16 stdin -> /proc/self/fd/0
lrwxrwxrwx 1 root root 15 May 30 18:16 stdout -> /proc/self/fd/1
### mount >/dev/console
none on /run type tmpfs (rw,noexec,
none on /dev type tmpfs (rw)
devpts on /dev/pts type devpts (rw,gid=
none on /dev/shm type tmpfs (rw)
FATAL: Module blcr not found.
FATAL: Module dm_mod not found.
[ 47.290884] init: Failed to create pty - disabling logging for job: 2
[ 47.304168] init: Failed to create pty - disabling logging for job: 2
...
So both /dev/ptmx is there and /dev/pts is mounted so from then on logging should work. But the errors persist.
Booting continues and at some point I get a shell:
beo-01:/# start test
[ 587.266601] init: Failed to create pty - disabling logging for job: 2
test
test stop/waiting
I'm booting a freshly bootstraped Precise with nfs-root. For that to work I had to replace mountall so the system isn't 100% pristine. I'm also using a custom kernel and no initramfs. But I've added some debug outputs to /dev/console there to see if everything works:
[ 0.000000] Linux version 2.6.32.
[ 0.000000] Command line: root=/dev/nfs rw nfsroot=
...
[ 38.048148] Freeing unused kernel memory: 528k freed
### after my own mount-virtfs job instead mountall:
### ls -lh /dev >/dev/console
total 0
crw------- 1 root root 5, 1 May 30 18:16 console
lrwxrwxrwx 1 root root 11 May 30 18:16 core -> /proc/kcore
lrwxrwxrwx 1 root root 13 May 30 18:16 fd -> /proc/self/fd
brw------- 1 root root 7, 0 May 30 18:16 loop0
drwxr-xr-x 2 root root 40 May 30 18:16 net
crw-rw-rw- 1 root root 1, 3 May 30 18:16 null
crw------- 1 root root 108, 0 May 30 18:16 ppp
crw-rw-rw- 1 root tty 5, 2 May 30 18:16 ptmx
drwxr-xr-x 2 root root 0 May 30 18:16 pts
drwxrwxrwt 2 root root 40 May 30 18:16 shm
lrwxrwxrwx 1 root root 15 May 30 18:16 stderr -> /proc/self/fd/2
lrwxrwxrwx 1 root root 15 May 30 18:16 stdin -> /proc/self/fd/0
lrwxrwxrwx 1 root root 15 May 30 18:16 stdout -> /proc/self/fd/1
### mount >/dev/console
none on /run type tmpfs (rw,noexec,
none on /dev type tmpfs (rw)
devpts on /dev/pts type devpts (rw,gid=
none on /dev/shm type tmpfs (rw)
FATAL: Module blcr not found.
FATAL: Module dm_mod not found.
[ 47.290884] init: Failed to create pty - disabling logging for job: 2
[ 47.304168] init: Failed to create pty - disabling logging for job: 2
...
So both /dev/ptmx is there and /dev/pts is mounted so from then on logging should work. But the errors persist.
Booting continues and at some point I get a shell:
beo-01:/# start test
[ 587.266601] init: Failed to create pty - disabling logging for job: 2
test stop/waiting
beo-01:/# telinit u
beo-01:/# telinit q
beo-01:/# mv /etc/init/test.conf .
beo-01:/# mv test.conf /etc/init
beo-01:/# start test
test stop/waiting
And that is where I get stuck. Why does upstart need to be restarted to be able to open pseudo ptys?
Why doesn't it work right after /dev and /dev/pts are created by the mount-virtfs job?
Can I restart upstart as part of the mount-virtfs job?
ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: upstart 0.6.5-8 [modified: sbin/initctl sbin/shutdown sbin/init sbin/telinit sbin/reboot sbin/upstart-
ProcVersionSign
Uname: Linux 2.6.35-23-server x86_64
Architecture: amd64
Date: Wed May 30 18:43:30 2012
ProcEnviron:
SHELL=/bin/bash
PATH=(custom, no user)
LANG=en_US.UTF-8
LANGUAGE=
SourcePackage: upstart