Hi,

thanks a lot for the discussion. I think Brian made clear what
completing the design doc regarding disks would mean. From my point of
view the "big can of worms" are only caused by DRBD-8.4's nature for
primary and secondary. All other (redundant) disk templates, which
Ganeti calls "externally mirrored" do not have such limitations.

It might be considered to not allow de-/attach for DRBD on the first
run, or even at all (same for file and plain?). Maybe the use case for
de-/attach itself is somewhat special. Completing the disk design doc,
would open the way for mixing disk types, again limited for the DRBD
case, but not for externally mirrored.

I also agree with Valentin, that solving this types of problems with
the external storage interface instead of inside Ganeti is not the most
valuable variant. Besides that, it can be observed, that this
"functions as executables" interfaces, namely OS and storage, tend to
everyone inventing their own wheel. Probably a subclassing and more
integrated approach would be better?

Valentin offered to contribute towards completing the disk design doc,
which would be a great improvement for the Ganeti project. Personally I
like to support this from my end. But this must been seen more from a
point of a longtime user, than a Ganeti developer (my programming
capabilities are enough for smaller fixes, but easily lost inside
Haskell). So I'm cc-ing to ganeti-devel here, probably to catch some
other's thoughts as well. The full discussion can be found at [1].

Thanks, Sascha.

[1] https://groups.google.com/g/ganeti/c/FwMZhBP9Ru0

-- 
You received this message because you are subscribed to the Google Groups 
"ganeti-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/ganeti-devel/20210215234128.b051a769b9434a06cfb93473%40web.de.

Reply via email to