LIO Target Synchronization with sysfs [resend]
[Apologies for the resend.]
I would like to attend LSF/MM this year. I would like to discuss problems I've had dealing with LIO targets and their inherent asynchronicity in sysfs. I believe that with judicious use of kernel events we can help user space handle target changes more cleanly.
As for other topics:
As one of the maintainers of LIO at SUSE, I'd dearly love to see our iSCSI targets merged.
I also would love to see multipath work in a sane way, and *with* the SCSI subsystem, not on top of it, so I'd like to join that discussion.
Signature · 1 line
Lee Duncan
Quoted 7 lines · Lee Duncan
[Apologies for the resend.]
I would like to attend LSF/MM this year. I would like to discuss problems I've had dealing with LIO targets and their inherent asynchronicity in sysfs. I believe that with judicious use of kernel events we can help user space handle target changes more cleanly.
I assume you mean configfs, and not sysfs. ;)
Quoted 5 lines
As for other topics:
As one of the maintainers of LIO at SUSE, I'd dearly love to see our iSCSI targets merged.
You're talking about Mike's (CC'ed) original target_core_rbd.c driver, right..?
He's been working on porting the PR related changes to a generic interface that can be consumed by multiple backend using target_core_iblock.c.
Quoted 10 lines · Nicholas A. Bellinger
Quoted 7 lines · Lee Duncan
[Apologies for the resend.]
I would like to attend LSF/MM this year. I would like to discuss problems I've had dealing with LIO targets and their inherent asynchronicity in sysfs. I believe that with judicious use of kernel events we can help user space handle target changes more cleanly.
I assume you mean configfs, and not sysfs. ;)
Apologies. Yes.
Quoted 7 lines
Quoted 6 lines
As for other topics:
As one of the maintainers of LIO at SUSE, I'd dearly love to see our iSCSI targets merged.
Quoted 7 lines
You're talking about Mike's (CC'ed) original
target_core_rbd.cdriver, right..?He's been working on porting the PR related changes to a generic interface that can be consumed by multiple backend using
target_core_iblock.c.
Actually, no. I'm aware of that work because one of the groups I'm in is working in that area. And that is of interest to me to.
But I meant the scst/lio merge, one of the proposed LSF/MM topics.
Signature · 1 line
Lee Duncan