On Fri, Jul 11, 2014 at 12:56:53PM +0200, Benoît Canet wrote:
> The Friday 11 Jul 2014 à 11:01:22 (+0800), Liu Yuan wrote :
> > This patch adds single read pattern to quorum driver and quorum vote is
> > default
> > pattern.
> >
> > For now we do a quorum vote on all the reads, it is designed for unreliable
> > underlying storage such as non-redundant NFS to make sure data integrity at
> > the
> > cost of the read performance.
> >
> > For some use cases as following:
> >
> > VM
> > --------------
> > | |
> > v v
> > A B
> >
> > Both A and B has hardware raid storage to justify the data integrity on its
> > own.
> > So it would help performance if we do a single read instead of on all the
> > nodes.
> > Further, if we run VM on either of the storage node, we can make a local
> > read
> > request for better performance.
> >
> > This patch generalize the above 2 nodes case in the N nodes. That is,
> >
> > vm -> write to all the N nodes, read just one of them. If single read
> > fails, we
> > try to read next node in FIFO order specified by the startup command.
> >
> > The 2 nodes case is very similar to DRBD[1] though lack of auto-sync
> > functionality in the single device/node failure for now. But compared with
> > DRBD
> > we still have some advantages over it:
> >
> > - Suppose we have 20 VMs running on one(assume A) of 2 nodes' DRBD backed
> > storage. And if A crashes, we need to restart all the VMs on node B. But for
> > practice case, we can't because B might not have enough resources to setup
> > 20 VMs
> > at once. So if we run our 20 VMs with quorum driver, and scatter the
> > replicated
> > images over the data center, we can very likely restart 20 VMs without any
> > resource problem.
> >
> > After all, I think we can build a more powerful replicated image
> > functionality
> > on quorum and block jobs(block mirror) to meet various High Availibility
> > needs.
> >
> > E.g, Enable single read pattern on 2 children,
> >
> > -drive driver=quorum,children.0.file.filename=0.qcow2,\
> > children.1.file.filename=1.qcow2,read-pattern=single,vote-threshold=1
> >
> > [1] http://en.wikipedia.org/wiki/Distributed_Replicated_Block_Device
> >
> > Cc: Benoit Canet <ben...@irqsave.net>
> > Cc: Kevin Wolf <kw...@redhat.com>
> > Cc: Stefan Hajnoczi <stefa...@redhat.com>
> > Signed-off-by: Liu Yuan <namei.u...@gmail.com>
> > ---
> > block/quorum.c | 174
> > +++++++++++++++++++++++++++++++++++++++++----------------
> > 1 file changed, 125 insertions(+), 49 deletions(-)
> >
> > diff --git a/block/quorum.c b/block/quorum.c
> > index d5ee9c0..2f18755 100644
> > --- a/block/quorum.c
> > +++ b/block/quorum.c
> > @@ -24,6 +24,7 @@
> > #define QUORUM_OPT_VOTE_THRESHOLD "vote-threshold"
> > #define QUORUM_OPT_BLKVERIFY "blkverify"
> > #define QUORUM_OPT_REWRITE "rewrite-corrupted"
> > +#define QUORUM_OPT_READ_PATTERN "read-pattern"
> >
> > /* This union holds a vote hash value */
> > typedef union QuorumVoteValue {
> > @@ -74,6 +75,16 @@ typedef struct BDRVQuorumState {
> > bool rewrite_corrupted;/* true if the driver must rewrite-on-read
> > corrupted
> > * block if Quorum is reached.
> > */
> > +
> > +#define READ_PATTERN_QUORUM 0 /* default */
> > +#define READ_PATTERN_SINGLE 1
>
> Why not making these choices an enum ?
It is just a matter of taste. For now we only have 2 constants and the bonus to
have #define here is we can put them near the 'read_pattern' for easier code
read.
> Would READ_PATTERN_SINGLE be more accuratelly described by
> READ_PATTERN_ROUND_ROBIN ?
How about FIFO? Since 'quorum' is a single word...
> More generaly would s/single/round-robin/ be more descriptive.
ditto.
>
> > + int read_pattern; /* single: read a single child and try first one
> > + * first. If error, try next child in an
> > + * FIFO order specifed by command line.
> > + * Return error if no child read
> > succeeds.
> > + * quorum: read all the children and do a quorum
> > + * vote on reads.
> > + */
> > } BDRVQuorumState;
> >
> > typedef struct QuorumAIOCB QuorumAIOCB;
> > @@ -117,6 +128,7 @@ struct QuorumAIOCB {
> >
> > bool is_read;
> > int vote_ret;
> > + int child_iter; /* which child to read in single pattern */
> > };
> >
> > static bool quorum_vote(QuorumAIOCB *acb);
> > @@ -256,6 +268,21 @@ static void quorum_rewrite_aio_cb(void *opaque, int
> > ret)
> > quorum_aio_finalize(acb);
> > }
> >
> > +static BlockDriverAIOCB *read_single_child(QuorumAIOCB *acb);
> > +
> > +static void quorum_copy_qiov(QEMUIOVector *dest, QEMUIOVector *source)
> > +{
> > + int i;
> > + assert(dest->niov == source->niov);
> > + assert(dest->size == source->size);
> > + for (i = 0; i < source->niov; i++) {
> > + assert(dest->iov[i].iov_len == source->iov[i].iov_len);
> > + memcpy(dest->iov[i].iov_base,
> > + source->iov[i].iov_base,
> > + source->iov[i].iov_len);
> > + }
> > +}
>
> > +
> > static void quorum_aio_cb(void *opaque, int ret)
> > {
> > QuorumChildRequest *sacb = opaque;
> > @@ -263,6 +290,21 @@ static void quorum_aio_cb(void *opaque, int ret)
> > BDRVQuorumState *s = acb->common.bs->opaque;
> > bool rewrite = false;
> >
> > + if (acb->is_read && s->read_pattern == READ_PATTERN_SINGLE) {
> > + /* We try to read next child in FIFO order if we fail to read */
> > + if (ret < 0 && ++acb->child_iter < s->num_children) {
> > + read_single_child(acb);
>
> Here the previous failed read is never finalized/freed.
quorum_aio_finalize() will free all the reads. But yes, I need to modify
quorum_aio_finalize() to free exact reads we have done (all the failed read +
succeeded read)
> > + return;
> > + }
> > +
> > + if (ret == 0) {
> > + quorum_copy_qiov(acb->qiov, &acb->qcrs[acb->child_iter].qiov);
> > + }
> > + acb->vote_ret = ret;
> > + quorum_aio_finalize(acb);
> > + return;
> > + }
> > +
> > sacb->ret = ret;
> > acb->count++;
> > if (ret == 0) {
> > @@ -276,7 +318,6 @@ static void quorum_aio_cb(void *opaque, int ret)
> > return;
> > }
> >
> > - /* Do the vote on read */
>
> Why removing this comment ?
Hmmm, probably at then time I thought the name of quorum_vote() already desribed
it. But yes, I shouldn't touch the lines that I don't modify.
Thanks
Yuan
