Mac managed workstations constantly drop AFP connection

[baliset]

We have a lab of 30 new, fast iMacs. Our dual processor G4 server handles our netbooting, web, mail, Mac manager server, AFP, user home dirs and so on.
We run a fast, switched 100-speed network. We run 10.2.6 server. All our workstations are 9.2.2 running Mac Manager 2.2.1 client.

We have the infuriating situation where users log in and randomly experience server drop outs, immediately or 30 minutes down the track. Thus, the “The server <servername> has dropped out” or similar dialog comes up. The alias on the user’s desktop to their home folder does not access the user’s folder. The “Users” volume holding their directory is greyed out. The dropouts seem unrelated to general network loading or the activity of that user. The only alternative is for the user to log out (losing any work) and logging back in again.

Sometimes I even get the server disconnection message when the computer is at (or before!) the login screen. I *know* that the server is not too-heavily loaded, and should cope well with this level of traffic.

Is there some way to re-nice the server AFP process to give it more CPU time? Is this recommended or likely to do any good? Does renicing a process persist across reboots?

I’d be happy if I could even get some idea of what logs to check to diagnose why this is happening so constantly. Help! Thanks.

[EMH]

we’ve had the exact same problems your describing. we had users logging in and were unable to mount their home directories – once we logged in off a local account – say root – and then mounted a volume using AFP, logged out, then it worked fine. so it appeared to be a problem with the AFP service. we put a crontask in to run a script which would mount and unmount an AFP server volume ever so often. this has kinda solved the problem, but it still happens sometimes, and we’re still plagued with extremely slow network performance. it can sometimes take up to 15 minutes for a user to log in and mount their home folder – we have an extremely large network. the LDAP lookups don’t seem to be taking long, so we suspect AFP again.

Does anyone have any solution to this problem?

Thanks in advance.

[Author]

Posts