[Drbd-dev] [PATCH] DRBD 8.0; Make sure barrier is sent after last data block in epoch

Graham, Simon Simon.Graham at stratus.com
Tue Jan 29 20:40:10 CET 2008

> BTW, I did white space changes to them.

Sorry - I do try to remember to use the right conventions when editing
these files, but it's different from our internal standard and I
sometimes forget!

> From the other patch bomb I still intend to merge
> 0010-Ensure-that-state-updates-cannot-be-done-out-of-orde.patch
> Regarding the other patches from patch bomb #1, I wanted to
> write an summay, which patches where accepted, which are not
> yet applied for what reasons. ... But then I went to the
> "Open Source Meets Business" Conference and now I can not
> remember...
> Simon, from your point of view, what patches to we missed from
> patch bomb #1 ?

"Patch bomb #1" -- I like it! Here's the set that I think have not been
applied yet (with the git SHA1 ids of the commits in my git repo - I
think these come in the .patch files, right?), including the one you
mentioned above:

. One more place where stop-sync flag is not explicitely cleared when
starting resync (16896a5)
. Ensure that state updated cannot be done out of order whilst running
receive_state (914435c)
. Revert "Fixing "drbdadm pause-sync r0; drbdadm resume-sync r0""
. Revert "make resync more robust, don't reset find bit offset too
early." (cab2179)

(note on these last two -- the changes you committed actually cause the
sync to stall if paused and resumed quickly. I suspect that some part of
these is probably desired, but you mixed in the changes to the way
resync is restarted which cause the stall so I just backed them out in


More information about the drbd-dev mailing list