Frequently Asked Questions | Sep 21, 2023 - 11:09pm |
|
|
Frequently Asked Questions Installation, Configuration and Related System Configuration 
How do I set up master/slave database configuration for an existing setup?
Rate This FAQ
(Not yet rated)
|
Created On: 23 Sep 1998 3:19 pm Last Edited: 10 Jun 2009 2:42 pm |
Question |
|
|
What is the best way to add a new slave/mirror to a cluster?
or
What is the best way to bring an existing crashed slave/mirror server back up? |
Answer |
If you simply turn on distributed servers on all instances in a cluster, then the ENTIRE database of the master must be transferred over TCP/IP for a proper synch to occur. Not only can this take forever, but it will have an effect on performance of the master and in extreme situations could bring your site down.
You can use the webx.set file to make setting up master/slave configurations much faster.
Updated Instructions
If you are bring up a completely new slave, follow all of this.
If you are simply restoring a database into an old slave at the same IP, you may skip steps 3, 4 and 5.
- login to what will be the master server and configure it as a master (if not already). Do a sysop backup to get an intact webxdb.1, or shut down the server and then use a copy of webx.db directly.
- Install your slave instance and make sure it is running OK. Then shut it down and copy the master's webxdb.1 from step 1 into it and rename to webx.db.
- copy the webx.set file from the master's directory, and change the
"isMaster/isSlave" settings so "isMaster=0" and "isSlave=1", update the FastCGI access port (if applicable), as well as any relevant IP settings to be correct for the relevant slave.
- If you utilize enclosures, you must also copy your enclosures directory from the master to the slave.
- If you are using any web files (HTML) you will need to copy these over as well, reproducing the physical directory paths in your original HTML directory on the new slave.
- Start up the slave
- NOTE: New for Version 4.1+ Due to the new search logic, it will be necessary to manually kick off the search indexes on all new slaves in the cluster. After each slave comes up, click "Rebuild Search Index" in their respective control panel(s).
It will immediately do an incremental synch to the master, and should start
taking requests (from your "router", or round-robin DNS, or Cisco
LocalDirector etc).
|
|
|
|
|