Re: [Qemu-devel] [PATCH 0/9] major rework of drive-mirror

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

* Denis V. Lunev (den@openvz.org) wrote:
> Block commit of the active image to the backing store on a slow disk
> could never end. For example with the guest with the following loop
> inside
>     while true; do
>         dd bs=1k count=1 if=/dev/zero of=x
>     done
> running above slow storage could not complete the operation with a
> resonable amount of time:
>     virsh blockcommit rhel7 sda --active --shallow
>     virsh qemu-monitor-event
>     virsh qemu-monitor-command rhel7 \
>         '{"execute":"block-job-complete",\
>           "arguments":{"device":"drive-scsi0-0-0-0"} }'
>     virsh qemu-monitor-event
> Completion event is never received.
> 
> This problem could not be fixed easily with the current architecture. We
> should either prohibit guest writes (making dirty bitmap dirty) or switch
> to the sycnchronous scheme.
> 
> This series switches driver mirror to synch scheme. Actually we can have
> something more intelligent and switch to sync mirroring just after
> the first pass over the bitmap. Though this could be done relatively
> easily during discussion. The most difficult things are here.

Some random thoughts:
   a) People have been asking for post-copy for block storage, and this
solves the same problem a different way.
   b) I guess the other way is to turn on write-throttling - that would
also guarantee the rate at which writes happen.
   c) Don't you end up with something verymuch like what the colo block replication
guys have ended up with?

Dave

> The set also adds some performance improvements dealing with
> known-to-be-zero areas.
> 
> Signed-off-by: Denis V. Lunev <den@openvz.org>
> Reviewed-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
> CC: Stefan Hajnoczi <stefanha@redhat.com>
> CC: Fam Zheng <famz@redhat.com>
> CC: Kevin Wolf <kwolf@redhat.com>
> CC: Max Reitz <mreitz@redhat.com>
> CC: Jeff Cody <jcody@redhat.com>
> CC: Eric Blake <eblake@redhat.com>
> 
> Denis V. Lunev (9):
>   mirror: fix calling of blk_aio_pwritev/blk_aio_preadv
>   mirror: create mirror_dirty_init helper for mirror_run
>   mirror: optimize dirty bitmap filling in mirror_run a bit
>   mirror: efficiently zero out target
>   mirror: improve performance of mirroring of empty disk
>   block: pass qiov into before_write notifier
>   mirror: allow to save buffer for QEMUIOVector in MirrorOp
>   mirror: use synch scheme for drive mirror
>   mirror: replace bdrv_dirty_bitmap with plain hbitmap
> 
>  block/io.c                |  12 +-
>  block/mirror.c            | 290 +++++++++++++++++++++++++++++++++++-----------
>  include/block/block_int.h |   1 +
>  3 files changed, 229 insertions(+), 74 deletions(-)
> 
> -- 
> 2.5.0
> 
> 
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK