| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As per community consensus, we have decided to rename
nsr to jbr(Journal-Based-Replication). This is the patch
to rename the "nsr" code to "jbr"
Change-Id: Id2a9837f2ec4da89afc32438b91a1c302bb4104f
BUG: 1328043
Signed-off-by: Avra Sengupta <asengupt@redhat.com>
Reviewed-on: http://review.gluster.org/13899
Smoke: Gluster Build System <jenkins@build.gluster.com>
CentOS-regression: Gluster Build System <jenkins@build.gluster.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Write infra for nsr_server to not send a
CHILD_UP before it gets a CHILD_UP from a
quorum of it's children. Using the CHILD_UP
received in the nsr client translator from
the server, to decide the right time for
starting the I/Os
Change-Id: I9551638b306bdcbc6bae6aeda00316576ea832fe
Signed-off-by: Avra Sengupta <asengupt@redhat.com>
Reviewed-on: http://review.gluster.org/13623
CentOS-regression: Gluster Build System <jenkins@build.gluster.com>
Smoke: Gluster Build System <jenkins@build.gluster.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
|
| |
|
|
|
|
|
|
|
|
|
| |
Change-Id: I5f16a04e34a37e1f4cef38d75aff3479429473e3
Signed-off-by: Jeff Darcy <jdarcy@redhat.com>
Reviewed-on: http://review.gluster.org/13625
Tested-by: Avra Sengupta <asengupt@redhat.com>
Smoke: Gluster Build System <jenkins@build.gluster.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Avra Sengupta <asengupt@redhat.com>
CentOS-regression: Gluster Build System <jenkins@build.gluster.com>
|
|
|
The NSR-server with this patch, appoints the first node
on every replica subvolume, as the leader for that subvolume.
On receiving a 'write' fop, the leader first checks if there is
quorum in the replica subvolume to proceeed. In case there isn't
it fails with EROFS.
If there is quorum, the leader forwards the fop to the
followers. The followers on receiving the fop, perform the
operation, and based on the success or failure of the outcome
send a +ve or a -ve ack to the leader.
The leader after receiving acks from the followers performs a
quorum check of the acks, to see if it should even try to perform
the fop. If quorum is not being met, and the leader's outcome
wouldn't affect quorum, then it would send -ve ack to the client
without even performing the fop. If quorum is being met, the leader
will then try the fop on itself, and based on it's outcome perform
a quorum check of all the acks received (this time, including it's
own). Based on the result of the quorum check (irrespective of the
outcome on the leader), a +ve or -ve ack is send back to the client.
Change-Id: I860654b74c53e9b139b37dba43848e5504df6dce
Signed-off-by: Avra Sengupta <asengupt@redhat.com>
Reviewed-on: http://review.gluster.org/12705
Smoke: Gluster Build System <jenkins@build.gluster.com>
Tested-by: Jeff Darcy <jdarcy@redhat.com>
CentOS-regression: Gluster Build System <jenkins@build.gluster.com>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Jeff Darcy <jdarcy@redhat.com>
|
