AFP server Intel XServe – running Mac OS X Server OS v10.5.5 – clients running mac OS X v10.5.6 – centralized LDAP server for authentication (iPlanet with leopard schema) –
the issue/symptom (this has happened to us twice in last 4 weeks): AFP net home dir users got message saying they could not write/save to their home directory any longer. (most users had been logged in for 7 hours that day – less than 20 users connected) the AFP Server volume was still mounted on the client macs (I could see it from cmd line via ‘df’ – I could also cd to the home dir and run cmds like ‘ls’, ‘mkdir’, ‘touch’, etc – the AFP server and the end user macs were also still properly resolving names, uids, etc from LDAP (verified via dscl) – from the cmd line on the end user macs I could run cmds like mkdir and touch, but I could not write any data to those files or directories that I created – they were all Zero kb.
On the server (via local console or ssh) if I ‘su’ed to the end user accounts, then I could still write to the files in their homes – but the users connected to their homes over AFP could not. A reboot of the AFP server fixed the issue on the two occasions this has occurred. (uptime on the server before the reboot was only 10 days – ‘top’ did not indicate a run away AFP server process – load avg was low on server only about .15, it does not seem to happen over a fixed amount of time – my guess is something a user does triggers it -) We have a handful of other Intel Leopard XServes running 10.5.5 and 10.5.6 with similar number of net home users and we have not seen the issue occur on any of them. I found nothing in the system.log file on the server, nor in the AppleFileServiceAccess.log – the system.log files on end user macs have messages like: kernel[0]: AFP_VFS afpfs_vnop_strategy: Cached write on failed 13
Has anyone else experienced this same issue? ( i have since tried upgrading this server to 10.5.6 server – but I’m not sure if that will help at all in resolving this issue or if it might still come back)
An tips on other troubleshooting to try if it happens again – or what else to look for in system.logs ? Thanks in advance for any help.
.
.
Comments are closed