Is it possible to develop a failover procedure?

[galonga]

Hi

Since it appears that it is impossible using crappy M$ tecnology to implement a true cluster without getting horrible volume corruptions, we have to retort to good ol´failover techniques right?

If that´s the case, we are wondering if there´s some kind of script, development language, component or something like that to allow us to develop a software that would monitor one of the servers and as soon as it found out it went down CONNECT AUTOMATICALLY TO THE EXISTING VOLUME so as to continue processing normally?

Maybe the SDK will do the job? Any ideas? Thanks!

[Val (staff)]

Do you need implement this on the client side?
If yes - the MS iSCSI initiator's WMI interface could help you.

[galonga]

yes, it´s going to run on the client side.

basically it will connect to the volume should the server using it crashes, and use the volume accordingly.

the volume will be the type that cannot be shared, as to prevent being used by more than one initiator and therefore corruption will be (at least theoretically) gone

we will develop it ourselves, as the failover softwares we tried proved to be untrustworthy

[Val (staff)]

yes, it´s going to run on the client side.

basically it will connect to the volume should the server using it crashes, and use the volume accordingly.

the volume will be the type that cannot be shared, as to prevent being used by more than one initiator and therefore corruption will be (at least theoretically) gone

we will develop it ourselves, as the failover softwares we tried proved to be untrustworthy

Ok.

Please let us know about the results.