[sheepdog] [PATCH] multi-sheep cluster in one zookeeper cluster.

[Xu Yifeng]

On 2015年02月27日 15:10, Ruoyu wrote:
>
> On 2015年02月27日 13:41, Xu Yifeng wrote:
>>
>> On 2015年02月27日 12:04, Liu Yuan wrote:
>>> On Fri, Feb 27, 2015 at 11:51:22AM +0800, Yifeng Xu wrote:
>>>> Is this patch really necessary ? I think zookeeper C client support
>>>> chroot, if you read zookeeper.c source code, you will find it,
>>>> here is some code quoted:
>>>>
>>>> /**
>>>>   * Create a zookeeper handle associated with the given host and port.
>>>>   */
>>>> zhandle_t *zookeeper_init(const char *host, watcher_fn watcher,
>>>>    int recv_timeout, const clientid_t *clientid, void *context, int 
>>>> flags)
>>>> ...
>>>> //parse the host to get the chroot if
>>>>      //available
>>>>      index_chroot = strchr(host, '/');
>>>>      if (index_chroot) {
>>>>          zh->chroot = strdup(index_chroot);
>>>>          if (zh->chroot == NULL) {
>>>>              goto abort;
>>>>          }
>>>>
>>> Intresting. How we take advantage of this feature? Did you mean that 
>>> we can
>>> just append the cluster id on to "host" parameter?
>>>
>>> Thanks
>>> Yuan
>>>
>> Yes, we can.
> Thanks for you advice. I have a problem in using this feature.
> sheep exit with error when I try to start it with cluster option: -c 
> zookeeper:127.0.0.1:2181/default,timeout=10000
>
> The error message is as below.
>
> Feb 27 15:00:21   INFO [main] zk_init(1401) version 3.4.6, address 
> 127.0.0.1:2181/default, timeout 10000
> Feb 27 15:00:21   INFO [main] zk_init(1421) the negociated session 
> timeout is 10000
> Feb 27 15:00:21  ERROR [main] zk_queue_init(644) failed, path 
> /sheepdog, no node
> Feb 27 15:00:21  ERROR [main] main(954) failed to create sheepdog cluster
>
> Is something wrong?

Make sure you have created a directory before starting sheepdog, on my 
machine, I initialized it by following commands:
/usr/share/zookeeper/bin/zkCli.sh
  create /default myfoo
  quit
sheep -czookeeper:um37.sa.cl:2181/default,timeout=10000 /home/xyf/sheepdata/