[j-nsp] management daemon of M20(9.4R4.5) is not running

[Martin T]

I have a Juniper M20 with Junos 9.4R4.5, which all of the sudden
doesn't support SSH login:


martint at martin:~> ssh 192.168.1.254

Enter passphrase for key '/home/martin/.ssh/id_dsa':
--- JUNOS 9.4R4.5 built 2009-11-16 16:23:14 UTC
could not open user interface connection: management daemon not running
Retry connection attempts ? [yes,no] (yes)
could not open user interface connection: management daemon not running
Retry connection attempts ? [yes,no] (yes)
could not open user interface connection: management daemon not running
Retry connection attempts ? [yes,no] (yes)
could not open user interface connection: management daemon not running
Retry connection attempts ? [yes,no] (yes)
could not open user interface connection: management daemon not running
Retry connection attempts ? [yes,no] (yes) no
Connection to 192.168.1.254 closed.
martint at martin:~>


Over console line I can see, that mgd is actually running:

root    1169  0.0  1.0 28948 21668  ??  I    25Feb11   0:54.90 /usr/sbin/mgd -N


I tried to execute another instance of mgd with "/usr/sbin/mgd -N",
but this resulted with:

"mgd: error: daemon MGD detects existing daemon using lock file"

As it turned out, mgd uses lock file in order to ensure, that only one
instance of mgd is running at any specific time. Then I tried to force
mgd to reload gently using "kill -HUP 1169", but it looks like mgd
ignores the hang up signal. Then I sent SIGTERM with "kill 1169",
which caused mgd to restart:

root 79861  3.1  1.0 24824 21124  ??  S    10:26PM   0:00.39 /usr/sbin/mgd -N

..but there is still no access to the router over SSH. However, now I
don't get the "could not open user interface connection" error.
Instead the SSH session just hangs there forever.

Any ideas, how to regain access to this machine over SSH? Or is the
reboot only option out there in such case?


regards,
martin