lammert
msg:4290522 | 3:17 am on Apr 1, 2011 (gmt 0) |
Hi dirtyweasel,
Plugging drives out of the production server and plugging them in the development server implies that you need at least the same type of hardware. But even then, most RAID controllers store information like serial numbers or cached data of connected drives in their memory, and a disk swap may be well detected as a hardware failure and trigger a reprovisioning of that disk. In your situation temporarily mounting one disk into the directory of the other server over the network may be a saver way to solve your issue. It is also more flexible as you can work at different versions of your application in different directories at the same time, and switch from one to another version on your development or production server by just changing the mounting point.
|
Maurice
msg:4293078 | 11:44 am on Apr 6, 2011 (gmt 0) |
@dirtyweasel
this is how we did it for the BT Worldwide extranet back in the day.
we had.
1 Dev system
2 Test System (run as a technical copy)
3 Live
We had identical hardware for all 3 systems (3x 4u sun boxes) - the main networks/tech guy joked he wanted all the parts from the same production batch.
We would have liked a fourth machine as a hot spare in the exchange as to replace a hardware failure we would have had to fly the test machine up. And I suspect we might have had to threaten BA or Easy Jet with being black listed if they refused the server a ticket.
We could test then upgrade the live by swapping drives from test to live (after changing a few config lines to change the domain mostly)
Do you really need hot swap its going to be expensive and you do need a separate machine just for testing before going down hot swap raid arrays route.
|
dirtyweasel
msg:4293124 | 1:49 pm on Apr 6, 2011 (gmt 0) |
switch from one to another version on your development or production server by just changing the mounting point.
Interesting idea. I'm trying to wrap my head around this. Tell me more.
|
|
