Re: [PATCH rfc 2/2] fabrics: allow user to pass hostname instead of traddr

[Sagi Grimberg <sagi@grimberg.me>]

>> Some users would like to use well known hostnames instead
>> of remembering ip addresses. So, allow users to pass --hostname
>> and we will attempt to resolve against the DNS.
>>
>> Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
>> ---
>>   fabrics.c | 75 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 75 insertions(+)
>>
>> diff --git a/fabrics.c b/fabrics.c
>> index a7d628b1f0c9..7bd95c4b0d10 100644
>> --- a/fabrics.c
>> +++ b/fabrics.c
>> @@ -33,6 +33,10 @@
>>   #include <sys/stat.h>
>>   #include <stddef.h>
>>   
>> +#include <sys/types.h>
>> +#include <arpa/inet.h>
>> +#include <netdb.h>
>> +
>>   #include "util/parser.h"
>>   #include "nvme-ioctl.h"
>>   #include "nvme-status.h"
>> @@ -60,6 +64,7 @@ static struct config {
>>   	char *nqn;
>>   	char *transport;
>>   	char *traddr;
>> +	char *hostname;
>>   	char *trsvcid;
>>   	char *host_traddr;
>>   	char *hostnqn;
>> @@ -857,6 +862,54 @@ static int build_options(char *argstr, int max_len, bool discover)
>>   	return 0;
>>   }
> 
> The code looks fine.
> 
> I realize 'hostname' in this context is referring to the remote host and
> resolves the target's transport address, but it just sounds potentially
> confusing given we have "host_traddr". I'm not sure of a better name,
> though, so if this is normal convention, then okay.

We could do it like ssh that can accept either IP or hostname (we check
if string is IP and if not, resort to name resolution)

Is that preferable?

_______________________________________________
linux-nvme mailing list
linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme