Quote:
|
Originally Posted by new_techie I have 2 servers that I connect to using RDP. To make things go faster I do half my work through one and half through the other. I'm wondering, if one of them fails or loses it's connection then how can I make it auto switch over? Say for example I'm working on Server 1, it fails, so I want it to automatically detect that Server 1 is no longer connected and would like it to switch over to Server 2 (and vice versa) is this possible anyone? |
The UNIX way is to have a pool of processors in a cluster, then when one fails the jobs just don't get dispatched to it anymore. There's various software projects providing PVM for example.
If you're a developer, take a look at dist-cc which farms out compiles, and use the parallel option of make (-j 3) works well on dual CPU box.
If you tell UNIX, you want to connect to a desktop, then you get connected to a particular machines desktop. It'd be the RDP client which you configure to change sessions on a server list. A sys admin would be pretty hacked off if that client switched machines on a failure, so he didn't notice the server falling over.
If you're a developer then distributing out the jobs yourself, is time consuming, you're probably not a very good OS scheduler.
Finally unless the nodes are configured to fail-over, they won't take over each other's IP address, and anyway it'd be insecure to let any old machine hijack a TCP/IP session, so this stuff tends to work with protocols that re-connect (HTTP) or use UDP packets (DOMAIN, NTP).