glusterfs-snapshot.git/rpc/rpc-transport, branch development

glusterd/snapshot: Addressed upstream review comments

2014-04-07T07:51:03+00:00

Code cleanup and review comments fixed.

Change-Id: Ie51e3a2f995295632cb1db05bab8b38603ab9a0a
Signed-off-by: Rajesh Joseph 
Reviewed-on: http://review.gluster.org/7399
Reviewed-by: Vijaikumar Mallikarjuna 
Reviewed-by: Avra Sengupta

protocol/server: copy the response into the payload instead of using the pointer

2014-01-13T04:53:39+00:00

The response structure filled up in server_submit_reply is local to the
function (i.e stack allocated) whose address is stored in the barrier
payload to use later while sending the replies. But after the function
is exited (server_submit_reply) the pointer is not valid and contains
invalid data, which either leads in a segfault due to illegal memory
access or reply not being sent as the total length of the reply obtained
from that memory might not be valid.

So instead of saving the pointer inside the payload, save the complete
reply itself.

Change-Id: I7d0b7b181584865199357a67165b99bf35def5ab
Signed-off-by: Raghavendra Bhat

rpcsvc: implement per-client RPC throttling

2013-10-28T07:33:19+00:00

Implement a limit on the total number of outstanding RPC requests
from a given cient. Once the limit is reached the client socket
is removed from POLL-IN event polling.

Change-Id: I8071b8c89b78d02e830e6af5a540308199d6bdcd
BUG: 1008301
Signed-off-by: Anand Avati 
Reviewed-on: http://review.gluster.org/6114
Reviewed-by: Santosh Pradhan 
Reviewed-by: Rajesh Joseph 
Reviewed-by: Harshavardhana 
Reviewed-by: Vijay Bellur 
Tested-by: Gluster Build System

transport/socket: don't try to set TCP_DELAY on unix domain sockets

2013-10-12T07:42:12+00:00

Change-Id: I290cd983bd0dff2e32e5ee90a12e888a3b31c6fd
BUG: 969461
Signed-off-by: Raghavendra G 
Reviewed-on: http://review.gluster.org/5954
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

core: block unused signals in created threads

2013-09-25T08:33:16+00:00

Block all signal except those which are set for explicit handling
in glusterfs_signals_setup(). Since thread spawning code in
libglusterfs and xlators can get called from application threads
when used through libgfapi, it is necessary to do this blocking.

Change-Id: Ia320f80521a83d2edcda50b9ad414583a0175281
BUG: 1011662
Signed-off-by: Anand Avati 
Reviewed-on: http://review.gluster.org/5995
Tested-by: Gluster Build System 
Reviewed-by: Amar Tumballi 
Reviewed-by: Vijay Bellur

glusterfsd: Round robin DNS should not be relied upon with

2013-09-06T22:56:34+00:00

config service availability for clients.

Backupvolfile server as it stands is slow and prone to errors
with mount script and its combination with RRDNS. Instead in
theory it should use all the available nodes in 'trusted pool'
by default (Right now we don't have a mechanism in place for
this)

Nevertheless this patch provides a scenario where a list of
volfile-server can be provided on command as shown below

-----------------------------------------------------------------
$ glusterfs -s server1 .. -s serverN --volfile-id= \
      
-----------------------------------------------------------------
                   OR
-----------------------------------------------------------------
$ mount -t glusterfs -obackup-volfile-servers=: \
      :...: :/ 
-----------------------------------------------------------------

Here ':' is used as a separator for mount script parsing

Now these will be remembered and recursively attempted for
fetching vol-file until exhausted. This would ensure that the
clients get 'volume' configs in a consistent manner avoiding the
need to poll through RRDNS.

Change-Id: If808bb8a52e6034c61574cdae3ac4e7e83513a40
BUG: 986429
Signed-off-by: Harshavardhana 
Reviewed-on: http://review.gluster.org/5400
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

mgmt/glusterd, socket: Change logging for brick disconnects

2013-06-11T14:52:16+00:00

For unix path based sockets, the socket path is
cryptic (md5sum of path) and may not be useful for
the user in debugging so log it in DEBUG.

Changed logging in brick_rpc_notify to log brickinfo
for disconnects.

Change-Id: I69174bbbbde8352d38837723e950ad8fc15232aa
BUG: 963153
Signed-off-by: Pranith Kumar K 
Reviewed-on: http://review.gluster.org/5009
Reviewed-by: Niels de Vos 
Reviewed-by: Krishnan Parthasarathi 
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

transport/socket: fix connect/disconnect races

2013-06-04T22:37:54+00:00

We might receive a connect request while a disconnect is still in
progress, requiring more states and (the return of) poller generation
numbers to avoid redundant pollers.  We might also get either kind of
request from within our own rpc_transport_notify upcall, so we have to
avoid locking and use the PLEASE_DIE state instead.

Change-Id: Icbaacf96c516b607a79ff62c90b74d42b241780f
BUG: 970194
Signed-off-by: Jeff Darcy 
Reviewed-on: http://review.gluster.org/5137
Tested-by: Gluster Build System 
Reviewed-by: Anand Avati

socket: Prevent extra logging in socket_rwv

2013-05-31T18:22:24+00:00

Problem:
If the socket is not even connected socket_rwv fails
and the error log comes every 3 seconds for every
re-connect.

Fix:
Prevent this error log if the socket is not even
connected.

Change-Id: I963f1345c4b6779dd5491948f7f5c4357a507b67
BUG: 963088
Signed-off-by: Pranith Kumar K 
Reviewed-on: http://review.gluster.org/5008
Reviewed-by: Amar Tumballi 
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur

socket: convert socket functions to file scope (static)

2013-05-29T12:26:57+00:00

The function names used in socket.c conflict with Samba's
socket functions (samba.git/source4/lib/socket/socket.h). Convert
our functions to file scope as they are anyways not called across
files.

Change-Id: If98ae557d3e2868f257c021b283ede6a5e92de02
BUG: 953694
Signed-off-by: Anand Avati 
Reviewed-on: http://review.gluster.org/5104
Tested-by: Gluster Build System 
Reviewed-by: Vijay Bellur