tch, csh, ksh hang

[lpb2ha]

Hi, we first experienced experienced this problem a few weeks ago. All of a sudden, on 3 different macs (Mac Pro 10.5.4 OS X Server, Mac Pro 10.5.4, and 2004 xserve, 10.4.* OS X Server), within the space of a few days, the 3 machines' were incapable of launching a shell.

1. When I try to log in to the machine through ssh remotely, after receiving my password, the terminal hangs.
2. Terminal hangs on launch
3. iTerm hangs on launch
4. xterm hangs on launch
5. Double clicking on tcsh, csh or ksh from Finder in /bin launches a Terminal window that hangs.

In all cases, I would not get a login prompt.

I tried from different users accounts that have different default shells (bash vs. tcsh). I tried changing the login shell from default to tcsh to zsh.

This I find to be very bizarre. I don't know what the commonality is for those applications. I don't think there is a common .login script that is used.

The computers otherwise seem to be working fine. I have web services, mysql, applications all seeming to work normally.

I can't see anything suspicious in my Console log files.

Last time I solved the problem on the 3 compouters by rebooting. But it has happened again (2-3 weeks later). This time I am tempted to let it be for a while until I come up with a possible solution that I can try (but not having remote access to the machine is not something that I can sustain.)

Anyone have a clue?

[Giaguara]

What do those remote computers show at console logs for the time the ssh hung?
What does top show?

[lpb2ha]

Thanks for asking.
I don't see much in my System.log on the client side. Terminal tells me it has launched a shell:
Aug 29 16:23:10 LBMBP login[41674]: USER_PROCESS: 41674 ttys002
but that is before I ssh.
The server responds to my login with some text that I will include with my next post (so I can copy and paste from the server), which indicates to me that the login attempt was in fact successful. I'm sure all is well from the client side since I get the same problems if I launch on the server itself.

I don't notice anything special about tops (activity monitor) on the server. One particularlity maybe is that I am running VNC (only accepts connections via ssh). Also, there is more than one sshd for some reason that I ignore. I'll paste the activity monitor output from the other machine in a few minutes.

[Mikuro]

It sounds a bit like this problem: Using Terminal . Short version: Try using Onyx to run the cron scripts and clear your font cache.

[lpb2ha]

server side system log when I log in from remote host to the server via ssh (which does not give me back a login prompt)

Aug 29 16:23:28 www com.apple.SecurityServer[55]: checkpw() succeeded, creating credential for user lucb
Aug 29 16:23:28 www com.apple.SecurityServer[55]: checkpw() succeeded, creating shared credential for user lucb
Aug 29 16:23:28 www com.apple.SecurityServer[55]: Succeeded authorizing right system.login.tty by client /usr/sbin/sshd for authorization created by /usr/sbin/sshd.
Aug 29 16:23:28 www sshd[69280]: Accepted keyboard-interactive/pam for lucb from 142.58.54.15 port 60614 ssh2
Aug 29 16:24:14 www sshd[69311]: /etc/sshd_config line 70: Unsupported option KerberosGetAFSToken
Aug 29 16:26:14 www sshd[69315]: error: ssh_msg_send: write

And the server's activity monitor:
69361 Activity Monitor admin 1.8 7 11.18 MB 976.68 MB Intel
201 AirPort Base Station Agent admin 0.0 1 2.88 MB 890.56 MB Intel
82616 AppleSpell.service admin 0.0 1 1.60 MB 601.20 MB Intel
210 AppleVNCServer admin 0.0 4 2.27 MB 850.67 MB Intel
213 ATSServer admin 0.0 2 5.38 MB 641.68 MB Intel
32398 AutoUpdate admin 3.3 2 27.23 MB 1.04 GB PowerPC
68226 Console admin 0.0 6 10.16 MB 957.75 MB Intel
90842 DiskImages UI Agent admin 0.0 1 6.70 MB 925.44 MB Intel
90840 diskimages-helpe admin 0.1 4 7.20 MB 605.59 MB Intel
214 Dock admin 0.0 3 17.98 MB 943.20 MB Intel
217 Finder admin 0.0 16 20.93 MB 978.03 MB Intel
90985 Firefox admin 4.5 13 88.70 MB 1,014.30 MB Intel
196 launchd admin 0.0 3 556.00 KB 585.74 MB Intel
85 loginwindow admin 0.0 3 6.87 MB 926.61 MB Intel
69327 mdworker admin 0.0 4 2.06 MB 598.96 MB Intel
8129 MySQL Query Browser admin 0.0 2 6.67 MB 928.00 MB Intel
90687 OSXvnc-server (Not Responding) admin 0.0 4 3.54 MB 877.05 MB Intel
209 pboard admin 0.0 1 588.00 KB 586.62 MB Intel
69356 Quick Look Server admin 0.0 5 2.57 MB 855.50 MB Intel
82599 Safari admin 0.0 10 71.89 MB 1,006.30 MB Intel
68174 smbclient admin 0.0 1 620.00 KB 586.83 MB Intel
68173 smbclient admin 0.0 1 620.00 KB 586.83 MB Intel
207 Spotlight admin 0.0 4 3.98 MB 861.00 MB Intel
66119 sshd admin 0.0 1 528.00 KB 589.65 MB Intel
66091 sshd admin 0.0 1 528.00 KB 589.65 MB Intel
67962 sshd admin 0.0 1 516.00 KB 589.65 MB Intel
69120 sshd admin 0.0 1 528.00 KB 589.65 MB Intel
7571 System Preferences admin 0.0 6 26.60 MB 980.33 MB Intel
216 SystemUIServer admin 0.1 9 8.14 MB 926.51 MB Intel
231 Terminal admin 0.0 6 11.04 MB 929.75 MB Intel
208 UserEventAgent admin 0.0 2 2.21 MB 844.75 MB Intel
3805 Vine Server admin 0.0 5 8.82 MB 928.67 MB Intel
36605 Workgroup Manager admin 0.0 5 13.46 MB 938.85 MB Intel

[lpb2ha]

Originally Posted by Mikuro

It sounds a bit like this problem: Using Terminal . Short version: Try using Onyx to run the cron scripts and clear your font cache.

I could try this. But I wouldn't know that this solved the problem, since previously I found that rebooting gave me a few weeks of respite. I don't think it's a font problem because I have the same problem from remote ssh. Also, xterm fonts I believe are different (at least they were in the old days). When you remote ssh, no font information is exchanged between the client and the server.

A colleague pointed me at the following thread from Apple:
http://discussions.apple.com/thread....sageID=7813879
in my case however, Terminal does not hang.

The problem started soon after we installed some OS X updates (10.5.4 and some 10.4 update I believe--i.e., we have a mixture of OS's, but as I recall rounds of patches became available at the same time). I strongly suspect an OS X bug. This would explain why 3 servers showed the same problem.