Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Clustering and file copy
- Thread starter Breezeman
- Start date
marvhuffaker
MIS
Blip is absolutely correct. But it depends on how you are copying files.. If you're using Windblows Exploder, it will most likely fail and abort, but that's a problem with the Windows design.. it's not a cluster problem.
Here's a short history of clustering that you need to understand...
If you were to run a 600MB Movie file off of your cluster resource.. say an AVI in Media Player on Windows XP. If the cluster resource were to failover - your movie file would be interrupted and it would start over from the beginning. This tells you that Windows is closing the file for you when the 'blip' occurrs.
Take that same movie file, run it on a Windows 98 box. Run it from the cluster just like before. Failover the cluster resource to the other node. Your movie will blip, but it will pick up where it left off as soon as the other cluster resource comes online. This proves that on Windows 98, the file isn't actually closed.
So this would prove to you that the cluster is doing it's job just fine, but Windows XP is actually what is causing the file copy to fail. There is a TID or Appnote that talks about this, but I'd have to dig it up.
When a cluster fails over, there is always a blip where the service will not be available. With a lot of services, you'd never notice this. Like, GroupWise - you shouldn't notice anything. iFolder shouldn't be affected. Web Services will hardly be noticeable. But doing a file copy, because the file is open, then the cluster resource deactivates off of one node and then reactivates - there will be a blip. This is true with other clustering technologies as well.
BTW, if you're serious about your cluster, you should put in a 3rd node.
Marvin Huffaker, MCNE
Here's a short history of clustering that you need to understand...
If you were to run a 600MB Movie file off of your cluster resource.. say an AVI in Media Player on Windows XP. If the cluster resource were to failover - your movie file would be interrupted and it would start over from the beginning. This tells you that Windows is closing the file for you when the 'blip' occurrs.
Take that same movie file, run it on a Windows 98 box. Run it from the cluster just like before. Failover the cluster resource to the other node. Your movie will blip, but it will pick up where it left off as soon as the other cluster resource comes online. This proves that on Windows 98, the file isn't actually closed.
So this would prove to you that the cluster is doing it's job just fine, but Windows XP is actually what is causing the file copy to fail. There is a TID or Appnote that talks about this, but I'd have to dig it up.
When a cluster fails over, there is always a blip where the service will not be available. With a lot of services, you'd never notice this. Like, GroupWise - you shouldn't notice anything. iFolder shouldn't be affected. Web Services will hardly be noticeable. But doing a file copy, because the file is open, then the cluster resource deactivates off of one node and then reactivates - there will be a blip. This is true with other clustering technologies as well.
BTW, if you're serious about your cluster, you should put in a 3rd node.
Marvin Huffaker, MCNE
- Thread starter
- #4
Okay, I get that a blip is in order. What about a complete loss of connection? I can't beleive that it is working as advertised. When doing a file copy I get the message that the network connection is no longer available. So what parameters either on the workstation or the cluster can I change to affect this?
HEY, WATCH THIS!
HEY, WATCH THIS!
- Thread starter
- #5
marvhuffaker
MIS
Realistically.. How often will you be doing a file transfer and have your cluster failover? Usually you do failover testing while you're tweaking the cluster, but in reality, you shouldn't be bouncing back and forth between Nodes just for kicks and giggles.
And if you think that sucks, put in a Microsoft cluster and learn the true definition of 'This Sucks'.
Marvin Huffaker, MCNE
And if you think that sucks, put in a Microsoft cluster and learn the true definition of 'This Sucks'.
Marvin Huffaker, MCNE
- Status
Similar threads
- Locked
- Question
- Locked
- Question