Os/x Server 10.4 Locks Out All Pc Fileshare Users Randomly

A

Appletosh

Registered
I just recently installed OS/X Server 10.4.2 on a G4 450mhz/512Mb system with a single 120GB IDE system disk and a RAID1 200GB IDE disk pair on a third-party PCI IDE controller.

The system configured easily, and adding user accounts went smoothly. Client computers are mostly PCs running Win98SE and WinXP. Each user has their own personal file share mounted as drive Z:, and access-control-list controlled access to various company-wide file shares.

Everything will work fine for hours, then with no warning, none of the users can access the fileshares. This morning it happened, and when I got onsite, I could log into the GUI console, but got the pinwheel of death. Did a hard restart, system rebooted OK, and I was able to log in. All services worked fine until late afternoon, then suddenly, a repeat: All file services become unaccessible.

The error log, when examined this morning after the reboot, did not show anything of interest beyond the initial system startup. There were no repeated errors leading up to the probable time of freezing.

This is a major reliability problem, and is really irritating my customer. What areas should be examined for problems? What events can trigger this type of behavior? Is this a normally expected behavior that the Mac needs to be rebooted every 12 hours?

Please advise, thanks.
 
I have some more information to add to the original question: I performed the 10.4.3 update, forced the ethernet settings on the Macintosh G4 to 100FD, and also set the switch port to which the Macintosh is connected to 100FD also. I also had read somewhere that the screen sleep process could cause excessive CPU loading, so in Energy Saver I disabled screen sleep, and verified there is no CPU or disk sleep. Aside from that, no other changes. I left Activity Monitor running, and ran top from an SSH terminal session for about 36 hours, and the system seemed fine. However, at times, the free physical memory drops to 7MB, then rebounds to 275M, at times when I know there are no file server users. I am curious if there are any known DoS attacks or other vulnerabilities? It just doesn't seem to be a SMB problem, but more of a performance issue.
 
When the physical memory drops and then goes back up, which process is gobbling up the memory then releasing it? It sounds like some process has a bad memory leak somewhere...
 
Unfortunately, I was not closely watching the top display at the time I noticed the memory consumption spike. I just SSH'd in to the system again and see 251M free, and there is virtually nothing active going on, but 53 sleepers and 189 threads. In any event, the system has remained 'alive', which is a serious step forward. The previous times it froze, it took many minutes to even launch a terminal window from the console, which sounds more and more like a resource issue.
 
You must log in or register to reply here.
Back
Top