I can't login anymore to the server, neither via Remote Desktop nor ssh

I've had this problem reoccur, and we are using 10.5.3. So it looks like it wasn't fixed by 10.5.3.

Basically, our server stopped accepting passwords. No ssh, no AFP, no remote desktop.

I had to go to the machine itself, and tried to login, but the locked screen saver also accepted no passwords. The only thing I could do was reboot with the power button.

In the logs, the only thing I found that could possibly show something was up was a single line around the time of trying to log in,

ApplePasswordServer.Server.log:Jun 11 2008 10:32:58 KERBEROS-LOGIN-CHECK: no principal (F0AAE67B53943DD005AB94C3C6A980A0954CB6A9@YAMATO.INTERNAL)

No idea if that is related to the problem. But I couldn't find any other logs that seemed relevant.

Also of interest, we had just used the VPN 10 minutes before. Possibly just a coincidence, but I recall someone else saying they had this problem sometimes after using their VPN.

I just ran into this problem for the first time today. I'm not sure what caused it but I note the problem might stem from servermgrd. The first unexpected log entry in /var/log/system.log around the time of the problem was:

servermgrd\[116\]: * -\[WPSession createRecordWithType:attributes:inNode:error:\]: * Warning: the record of type dsRecTypeStandard:Config was created successfully, but OpenDirectoryFramework cannot find it. More info: recordName 'ServicesInformation1' in /Local/Default

It happened right after launchd started a new instance of a service (bootpd in this case), but it's probably unrelated to the service that starts up. After that servermgrd goes into a tail spin creating tons of ServicesInformation*.plist files in /var/db/dslocal/nodes/Default/config/ and other processes have problems looking up users and groups. I needed to get things working again, so I just followed the advice above, cleared out the ServicesInformation?*.plist files and rebooted. Next time this happens, I may try killing DirectoryServices and letting it restart to see if that fixes things.

I'm also on 10.5.3, so I agree the recent upgrade doesn't seem to have fixed this problem.

Arg. It happened again only 5 days after the last reboot. This doesn't bode well.

On a lark, I tried 'killall DirectoryServices' and it appeared to fix things, although with the weird state of things (looking at the mess in the system.log file), I think I'll reboot just to be on the safe side.

After your information I checked several servers.
I had one with this problem. In var/db/dslocal/nodes/Default/config/ there where a lot of ServicesInformation-files. Servers that did not have this problem have just one ServicesInformation-file.
The generated ServicesInformation-files are a lot smaller than the original.

My guess: When the server fails, he generates these files. I think after a launchd-script. Can anyone confirm that the problems start at 2.00 pm??

knn wrote:

Does anybody know some e-mail adrese that i can report to apple.

Go here: https://bugreport.apple.com

Log it as an actual bug. Others should do so also, so Apple gets some idea about just how big a problem this is for users of Leopard Server.

I'm seeing the same behavior on a Mac Pro 2 x 2.8 QuadXeon/OS X Server 10.5.3. I've actually set up 14 Leopard Servers this summer and this is the only one exhibiting this behavior. On top of that, after trying numerous things to fix it including updating to 10.5.4 Server and then reformatting and totally rebuilding the server, I actually moved that server to another campus and brought in a brand new server. I configured the brand new server from scratch (did not recover anything from a backup) and now the new server is showing the same behavior! So, that's two completely separate server boxes showing the same behavior on the same campus. All of my other 13 Leopard servers are humming along happily.

To add to this, I took the original server (which, by the way, is also brand new) and moved it to another site where I reformatted and rebuilt it from scratch (like I had tried on the troubled campus) and, so far, it's running just fine at the other site.

Sooooo, all of this leads me to believe that there must be some condition that exists (network or power?) at the one campus that's causing this strange behavior.

- s

I'm also having the same problem. After a few weeks it becomes impossible to access the server without restarting.

Really causing problems for me from a webhosting point of view.

If you can manually get to the server and a user is logged in already. Check the Activity Monitor. Whenever I do this all the processes have '????' in the user column. I guess this is why it cannot be accessed because the whole system seems to be running as ???? user?

Can anyone shed any light on this problem? It almost makes mac os x server useless for me 😟

A little update.

I mentioned earlier that I set up 14 Leopard Servers this summer and only one had this strange behavior. Well, now, I've seen it 6 of the 14 servers.

Not much else to report in the form of troubleshooting. I guess I'm waiting for a magical update.

- s

Folks

I am having the same issue. SSH and every other service can no longer login. If I am logged in at the time it happens and try to access the local accounts through the account preference screen, it is just comes up blank blank. Reboot clears up everything for 3- 5 days, before it comes back.

Vince

Having same problem here on our Open Directory master that is also serving as the PDC for windows machines on campus. So whenever it crops up (every 6 or 7 days on average) no windows box can log on at all.

I've spoken with premium support about this multiple times. the latest guy wanted me to wipe the server and reinstall. He told me my open directory was corrupted. I hesitated on doing so, since that means the OD archive tool wouldn't work (since it would archive the corruption) and exporting users out of workgroup manager would mean losing everyones passwords. Now I find this thread and learn it is not just me. so I'm pretty upset.

this machine is central to nearly everything on our network. I can't have it just up and stop responding to password authentication once a week. this is ridiculous.

I'd post to the bug report, but I've let my ADC acct lapse.