diff options
author | Lars Ellenberg <lars.ellenberg@linbit.com> | 2013-01-21 15:43:41 +0100 |
---|---|---|
committer | Philipp Reisner <philipp.reisner@linbit.com> | 2013-01-21 22:58:36 +0100 |
commit | 2681f7f6ce6c7416eb619d0fb19422bcc68bd9e1 (patch) | |
tree | 31b6a1a1830c4c81d917b545e659c1f46bcc1218 /drivers/block | |
parent | d2ec180c23a5a1bfe34d8638b0342a47c00cf70f (diff) | |
download | op-kernel-dev-2681f7f6ce6c7416eb619d0fb19422bcc68bd9e1.zip op-kernel-dev-2681f7f6ce6c7416eb619d0fb19422bcc68bd9e1.tar.gz |
drbd: fix potential protocol error and resulting disconnect/reconnect
When we notice a disk failure on the receiving side,
we stop sending it new incoming writes.
Depending on exact timing of various events, the same transfer log epoch
could end up containing both replicated (before we noticed the failure)
and local-only requests (after we noticed the failure).
The sanity checks in tl_release(), called when receiving a
P_BARRIER_ACK, check that the ack'ed transfer log epoch matches
the expected epoch, and the number of contained writes matches
the number of ack'ed writes.
In this case, they counted both replicated and local-only writes,
but the peer only acknowledges those it has seen. We get a mismatch,
resulting in a protocol error and disconnect/reconnect cycle.
Messages logged are
"BAD! BarrierAck #%u received with n_writes=%u, expected n_writes=%u!\n"
A similar issue can also be triggered when starting a resync while
having a healthy replication link, by invalidating one side, forcing a
full sync, or attaching to a diskless node.
Fix this by closing the current epoch if the state changes in a way
that would cause the replication intent of the next write.
Epochs now contain either only non-replicated,
or only replicated writes.
Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
Diffstat (limited to 'drivers/block')
-rw-r--r-- | drivers/block/drbd/drbd_req.c | 2 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_req.h | 1 | ||||
-rw-r--r-- | drivers/block/drbd/drbd_state.c | 7 |
3 files changed, 9 insertions, 1 deletions
diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index f58a4a4..2b8303a 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -168,7 +168,7 @@ static void wake_all_senders(struct drbd_tconn *tconn) { } /* must hold resource->req_lock */ -static void start_new_tl_epoch(struct drbd_tconn *tconn) +void start_new_tl_epoch(struct drbd_tconn *tconn) { /* no point closing an epoch, if it is empty, anyways. */ if (tconn->current_tle_writes == 0) diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h index 016de6b..c08d229 100644 --- a/drivers/block/drbd/drbd_req.h +++ b/drivers/block/drbd/drbd_req.h @@ -267,6 +267,7 @@ struct bio_and_error { int error; }; +extern void start_new_tl_epoch(struct drbd_tconn *tconn); extern void drbd_req_destroy(struct kref *kref); extern void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m); diff --git a/drivers/block/drbd/drbd_state.c b/drivers/block/drbd/drbd_state.c index 53bf618..0fe220c 100644 --- a/drivers/block/drbd/drbd_state.c +++ b/drivers/block/drbd/drbd_state.c @@ -931,6 +931,7 @@ __drbd_set_state(struct drbd_conf *mdev, union drbd_state ns, enum drbd_state_rv rv = SS_SUCCESS; enum sanitize_state_warnings ssw; struct after_state_chg_work *ascw; + bool did_remote, should_do_remote; os = drbd_read_state(mdev); @@ -981,11 +982,17 @@ __drbd_set_state(struct drbd_conf *mdev, union drbd_state ns, (os.disk != D_DISKLESS && ns.disk == D_DISKLESS)) atomic_inc(&mdev->local_cnt); + did_remote = drbd_should_do_remote(mdev->state); mdev->state.i = ns.i; + should_do_remote = drbd_should_do_remote(mdev->state); mdev->tconn->susp = ns.susp; mdev->tconn->susp_nod = ns.susp_nod; mdev->tconn->susp_fen = ns.susp_fen; + /* put replicated vs not-replicated requests in seperate epochs */ + if (did_remote != should_do_remote) + start_new_tl_epoch(mdev->tconn); + if (os.disk == D_ATTACHING && ns.disk >= D_NEGOTIATING) drbd_print_uuids(mdev, "attached to UUIDs"); |