Jump to content
Claris Engage 2025 - March 25-26 Austin Texas ×

This topic is 7970 days old. Please don't post here. Open a new topic instead.

Recommended Posts

Posted

This may have been a fluke, but I would prefer to be able to avoid it in the future.

FMP Server 5.5 on OS 9.04, 400 MHz iMac. 88 files. Peaks at about 30 users. At peak usage the cache hits look fine. Databases are speedy.

System has run for over a year without a crash (was running FM Server 3.0 until several months ago). I reboot the system roughly every two weeks.

Backs up self to hard drive at 12:15 (daytime).

Server gets backed up to DLT1 around 12:40 (daytime) with Retrospect.

Both of these operations slow things down for a couple minutes during the lunch hour, but at least we're definitely backed up.

A user with FM 6 (v3 upgrade) on Windows 98 claims that she couldn't get out of the squiggly arrow roughly in the middle of these backups. So she killed FM in her task manager, restarted it, and continued to work.

A few minutes later two users come to me saying that their FM is frozen. Mine is working fine. I address the two clients by killing the process and restarting it on them. I navigate a bit, do some searches, seems fine. Wait a minute. When the user tries to go to a particular record, it locks up (squiggly arrow) and doesn't recover.

I go to the server and take a closer look at the Administration window (that was the first place I looked when they reported the problem). There is a user logged in twice. And it shows her IP address twice. Yikes. Not good. Asked her to log off. Look at Admin window again. Now she's only on once. Asked her to reboot. It's still there, even when I try to disconnect the now phantom user. Note that this is all in the server's mind, so to speak. I cannot ping that IP because the user dutifully shut down their machine. So it's not a case of someone stealing an IP address.

My client is still working fine, so for kicks I try to go to the same record that crashed the other two users. Blammo. Squiggly arrows forever (until I kill the process).

Turns out that the Win98 user whose IP appeared twice in the Admin window was in that record when it froze. In the end I had to reboot the server to fix the problem.

I know that Mac OS 9 is supposedly not very stable, but this is literally the first problem I have had with this configuration.

Any suggestions of preventive measures? Thanks in advance.

-Chris

Posted

Well I back up on the fly at 1AM on OSX10.2. On classic Mac I ran an applescript that shut down the server, backed up and restarted the server at 1AM.

Posted

I should add that the noon backup is only one of four daily backups. The others are before most users get to work, after most go home, and once at 2am. The backups of the entire hard drive to tape are twice a day. My assumption is that the slowness during that particular noon backup, combined with a user killing FM on their client, probably combined with some bad timing, is what caused this to happen.

So the solution would seem to be not to run the noon backup. However, this problem has never happened before, and I really like the security of copying each morning's activity to the hard drive and to tape.

Posted

When the FMS backup routine runs, the file that is being backed up is temporarily suspended and any users connected to that file are also suspended - whatever they are doing is just put on hold until the file is finished and resumes. If a user Force-quits during this time, they are not properly logged off the server - and so a ghost IP address. You're also right that the only way to fix it is to reboot.

For some reason we have found that this only happens on FMS on Macs - FMS on Windows (NT4, W2K) seems to be able to resolve this issue. You do get an entry in the Event Log when a user looses connection.

I guess the options are: 1. forego the noontime backup routine, or 2. Make sure your users are aware of the lunchtime backup and to either logoff during lunch or not to force-quit if they get caught by the backup (squiggly cursor).

Posted

I just remembered another possible solution if you REALLY HAVE to do the back up during normal hours.

You could set up a separate (dedicated) workstation that could run a customized copy of the database, or even just a specialized opener file. This file could either run a constant script with timed pauses and launch at specific times. or

Setup the special file with a script set to run on opening and setup the Tash Scheduler to open the file at the desired time.

Either scenario would open the server file (you would have to set a default password to allow export) then export the data from the file to a filemaker file. You wouldn't have the layouts or scripts, but you would have the complete data set and the users would not get 'delayed' or 'squiggled'.

This topic is 7970 days old. Please don't post here. Open a new topic instead.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.