did you ever try ... when your main server #1 is down or unhealthy (no, not pig flue, filemaker flue :smileywink: ), disconnect it from ethernet, switch server #2 to ipadress #1 (and maybee servername #1 too), restart, and see if the clients see the files on server #2 on ip #1 without delays, and without any changes in your data files?
greetings from germany
I'm quite sure that if I changed the IP address of the back up server to that of the primary server it would eliminate the delays.
This business processes a very large volume of customer transactions in a very small amount of time. Even delays of a few minutes in serving them results in customers leaving or choosing not to stop when they see the parking lot is jammed.
I'm trying to set up a very fast "fail over" procedure that a non skilled user can use. Changing IP Addresses will take longer and require a skill set the user won't have. It's also complicated by a Business Owner/Manager who refuses to place all the files on Server 1, but instead wants half the files on Server 2 and Half on Server 1. I thus have set things up so that the two machines back each other up... So until I can convince him to host all files on a single server, changing the IP addresses will speed up file opens for one set and slow it down for the set already being hosted on the server.
Back to the original question: My tests show that just using the relevant path address, file: filename works without error. I'm concerned that some as yet undiscovered problem with this approach will ambush us. What problems might we encounter with our external file references set up this way?
hmmm, the only way for a fast switch i can see (presuming an "low skill administrator"), is a special admin login on the server. With a login script that leads the user to a menu, where he can choose between two options, and nothing else.
A) switch this machine to its first task, playing server #2
B) switch this machine to backup mode for #1
where each selection changes ip adress and reboots.
But ... having half the files on machine #1 and the other half on machine #2 might cause problems in this scenario. You will get in every emergency the other half of the files causing delays. So you should convince the business owner that having all files on one machine is best practice, with the second as backup only.
Or ... you can try to convince Filemaker to add a server mirroring feature in the next two weeks. (ehm, yesss, just kidding).
or ... you can try to use odbc connection to both machines to write a mirroring feature yourself (kidding again. ... but, not kidding ... it might work if you have no media fields in your database files).
greetings from germany
ps.: one last question ... blue user name, what does it mean? a count of posts you have reached, or filemaker employee?
I wondered about the blue color too, but have figured out that it is based on number of posts. Red marks a Filemaker employee such as TSGal.
I think you've mis-understood my question. I have a fully functional system and I can respond to server failure in minutes. Each of my files open rapidly whether they are hosted on server 1 or server 2. If server 1 fails, I go to server 2 and open the back up files that are uploaded to the server but remain closed except for emergency. If server 2 fails, I go to server 1 and do the same on it. Elapsed time starting from when I get the admin console open and listing the database file: less than a minute.
This method only works smoothly with all the file references based on File: filename instead of fmnet:/IP address/Filename.
Hence I repeat my question: Does anybody know of a reason why setting up these references like this would be a bad idea?
I found my own limitation with this approach which is probably obvious to many other Forum members:
If you only have relative path references to your files, they will successfully from the server if the files are in the same folder on the same server's hard drive but not if the referenced file is hosted from a different server. Thus, in a file on server 1, if I don't have a reference that uses a DNS name or IP address in the reference to a file that resides on server 2, any open file or perform external script reference will trigger a "file not found" error.
Set the references up in this format:
If a file is not found via relative path, the reference will resolve to the specified IP address and open the file with no noticeable delay. This allows me to open the file from the back up server without encountering unacceptable delays in getting the files to open.