The name of the configurable options should represent the services
provided or consumer of the port and not the protocol used.
* Renamed 'rpcaddress' to 'peeraddress': Denotes the address used
for peer to peer communication.
* Renamed 'restaddress' to 'clientaddress': Denotes the address
used by clients to communicate with glusterd2.
Coincidentally, these are analogous to those provided by etcd.
Signed-off-by: Prashanth Pai <ppai@redhat.com>
Peer addresses given during volume create may or may not have
a port number. This should be taken into account when check is
made if peer is present or not based on the address.
Signed-off-by: Prashanth Pai <ppai@redhat.com>
Barring transaction subpackage which contains mainly generated code,
other subpackages have had their golint warnings fixed.
We would want to ensure that all new code adheres to golint.
Signed-off-by: Prashanth Pai <ppai@redhat.com>
During review of daemon-framework PR, these functions were copied from
utils package to daemon package which consumes them. However, I missed
removing these from utils package. Doing that now.
Signed-off-by: Prashanth Pai <ppai@redhat.com>
Every transaction will now have an associated Context, that is passed to
all the steps. Anything that needs to be passed to/between steps can be
set in the transaction context.
Any logging done by the transaction framework during a transaction will
use the logger in the transaction Context.
Adding member upon peer probe in the etcd cluster.
With this change it will add remote host url during
peer probe of remote host in the cluster.
Issue #72
Signed-off-by: Gaurav Kumar Garg <garg.gaurav52@gmail.com>
Currently when etcd daemon start upon starting of glusterd2 then upon
performing ^C (SIGINT) its killing etcd daemon.
With this patch it will set child process id to different from main
process id group and, etcd daemon will not be kill upon performing
SIGINT (^C) of main process.
Addressing previous commit comment
Issue #66
Signed-off-by: Gaurav Kumar Garg <garg.gaurav52@gmail.com>
Signed-off-by : Atin Mukherjee <amukherj@redhat.com>
Currently when etcd daemon start upon starting of glusterd2 then upon
performing ^C (SIGINT) its killing etcd daemon.
With this patch it will set child process id to different from main
process id group and, etcd daemon will not be kill upon performing
SIGINT (^C) of main process.
Fixing some minor issue in unit text case and etcd restart case
Issue #66
Signed-off-by: Gaurav Kumar Garg <garg.gaurav52@gmail.com>
Signed-off-by : Atin Mukherjee <amukherj@redhat.com>
Currently when etcd daemon start upon starting of glusterd2 then upon
performing ^C (SIGINT) its killing etcd daemon.
With this patch it will set child process id to different from main
process id group and, etcd daemon will not be kill upon performing
SIGINT (^C) of main process.
Fixing previous commit comments.
Issue #66
Signed-off-by: Gaurav Kumar Garg <garg.gaurav52@gmail.com>
Signed-off-by : Atin Mukherjee <amukherj@redhat.com>
While creating volume using glusterd2 it should support creation of
hirearchical directory stracture, if brick present in hierarchical
directory.
Solving previous commit comment (adding unit test case)
Issue #68
Signed-off-by: Gaurav Kumar Garg <garg.gaurav52@gmail.com>
While creating volume using glusterd2 it should support creation of
hirearchical directory stracture, if brick present in hierarchical
directory.
Issue #68
Signed-off-by: Gaurav Kumar Garg <garg.gaurav52@gmail.com>
This patch reenables the volume package's unit test by mocking out few
dependencies like etcd, xattr calls.
Closes #57
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
This patch introduces the validation, commit & rollback phases (functions only)
of the volume create command life cycle. Later on all these functions can be
invoked by the transaction framework.
Closes #42
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
Please note currently the unit tests are going fail because of non-root
privileges. We'd need to move this infrastructure to something where we can have
root privileges
Signed-off-by: Atin Mukherjee <amukherj@redhat.com>
