Direct network link for glusterfs servers and client mount glusterfs from lan

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 06/25/2013 06:22 PM, Joe Julian wrote:
>
> On 06/25/2013 09:05 AM, Tamas Papp wrote:
>> On 06/25/2013 05:20 PM, Chiku wrote:
>>>
>>> Hello
>>>
>>> I have 2 gluster servers (10.0.1.11 and 10.0.1.12) and 1 client(10.0.1.1) on same lan.
>>> I added a direct network link between the 2 servers (10.10.10.11 and 10.10.10.12).
>>>
>>> From 1 gluster server :
>>>
>>> $ sudo gluster peer probe 10.10.10.12
>>>
>>> $ sudo gluster peer status
>>> Number of Peers: 1
>>>
>>> Hostname: 10.10.10.12
>>> Uuid: 467b70c7-c5b5-48c3-9026-4a2b3eadc5b8
>>> State: Peer in Cluster (Connected)
>>>
>>> $ sudo gluster volume create VOL_REPL2 replica 2 transport tcp 10.10.10.11:/opt/data/b2 
>>> 10.10.10.12:/opt/data/b2
>>>
>>> $ sudo gluster volume info
>>>
>>> Volume Name: VOL_REPL2
>>> Type: Replicate
>>> Volume ID: 65db00ee-5d64-490b-8f37-47d918f20062
>>> Status: Started
>>> Number of Bricks: 1 x 2 = 2
>>> Transport-type: tcp
>>> Bricks:
>>> Brick1: 10.10.10.11:/opt/data/b2
>>> Brick2: 10.10.10.12:/opt/data/b2
>>>
>>> From my client :
>>> $ sudo mount -t glusterfs 10.0.1.11:/VOL_REPL2 /mnt/VOL_REPL2
>>> Mount failed. Please check the log file for more details.
>>>
>>>
>>> $ sudo less /var/log/glusterfs/mnt-VOL_REPL1.log
>>> ...
>>> [2013-06-25 16:53:15.419615] E [socket.c:1715:socket_connect_finish] 0-VOL_REPL2-client-0: 
>>> connection to  failed (Connection refused)
>>> [2013-06-25 16:53:15.419781] E [socket.c:1715:socket_connect_finish] 0-VOL_REPL2-client-1: 
>>> connection to  failed (Connection refused)
>>> ...
>>>
>>> $ sudo mount -t nfs 10.0.1.11:/VOL_REPL2 /mnt/VOL_REPL2
>>> That works fine.
>>> But why I can't mount with glusterfs ?
>>
>> Try this on client:
>>
>> $ ip ro add 10.10.10.11 via 10.0.1.11
>> $ ip ro add 10.10.10.12 via 10.0.1.12
>> $ mount -t glusterfs 10.10.10.11:/VOL_REPL2 /mnt/VOL_REPL2
>>
> To expand on that a little, the native client always connects directly to the bricks as defined in 
> the volume definition. If you need a different route path, that's a networking problem at that 
> point, or you can use hostnames and split-horizon dns.

BTW, is it worth using a separated network for the cluster?

tamas


[Index of Archives]     [Gluster Development]     [Linux Filesytems Development]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux