Re: [Gluster-devel] problem with file locking use fcnt

[Dale Johnson]

Thanks, that did it.  I did need one extra

 option auth.ip.locks.allow *

for the new volume.

Is there a downside to just always running with this setting?  It seemed like
my timing tests writing files are pretty much the same, although I did use
large files to test with.

Dale.

On Fri, Jun 20, 2008 at 3:38 AM, Krishna Srinivas <address@hidden> wrote:
>> volume remote1
>>  type protocol/client
>>  option transport-type tcp/client
>>  option remote-host 2A
>>  option remote-subvolume brick
>> end-volume
>
> remote-subvolume should be locks, not brick. Same with the
> other client config.
>
> Regards
> Krishna
>
> On Fri, Jun 20, 2008 at 3:28 PM, Dale Johnson <address@hidden> wrote:
>>
>> First, gluster is a great product in a great space, and has been working 
>> well.
>> But I have hit a roadblock trying to do this:
>>
>> #include <stdio.h>
>> #include <fcntl.h>
>> #include <errno.h>
>> #include <unistd.h>
>> #include <string.h>
>>
>> int main( int argc, char ** argv ) {
>>  if( argc == 2 ) {
>>    int fd = open( argv[1], O_WRONLY | O_TRUNC | O_CREAT, 0666 );
>>    char s[] = "hello, world\n";
>>    write( fd, s, sizeof( s ) );
>>    close( fd );
>>
>>    int fd2 = open( argv[1], O_RDONLY );
>>    char buf[80];
>>    memset( buf, '\0', sizeof( buf ) );
>>    read( fd2, buf, sizeof( buf ) );
>>    printf( "buf: %s\n", buf );
>>
>>    struct flock lk;
>>    lk.l_type = F_RDLCK;
>>    lk.l_whence = SEEK_SET;
>>    lk.l_start = 0;
>>    lk.l_len = 6;
>>    lk.l_pid = 0;
>>
>>    int i = fcntl( fd, F_SETLK, &lk );
>>    printf( "i=%d,  errno=%d\n", i, errno );
>>    perror( "foo" );
>>
>>    close( fd2 );
>>  } else {
>>    printf( "usage: %s <outfile>\n", argv[0] );
>>    exit( 2 );
>>  }
>> }
>>
>> and I get this:
>>
>> address@hidden src]# ./a.out /mnt/glusterfs/fooX
>> buf: hello, world
>>
>> i=-1,  errno=38
>> foo: Function not implemented
>>
>> but not on a regular disk
>>
>> address@hidden src]# ./a.out /tmp/fooY
>> buf: hello, world
>>
>> i=0,  errno=0
>> foo: Success
>>
>> I would have expected this to work.  I'm running fuse 2.7.3,
>> glusterfs-1.3.9, and centos 2.6.18-53.1.21.el5 on intel.
>>
>> My config is 2 servers, and 2 clients (running under vmware), and the
>> server config is:
>>
>> volume brick
>>  type storage/posix
>>  option directory /data/export
>> end-volume
>>
>> volume locks
>>  type features/posix-locks
>>  option manditory on
>>  subvolumes brick
>> end-volume
>>
>> volume server
>>  type protocol/server
>>  option transport-type tcp/server
>>  option auth.ip.brick.allow *
>>  subvolumes brick
>> end-volume
>>
>> My client config is:
>>
>> volume remote1
>>  type protocol/client
>>  option transport-type tcp/client
>>  option remote-host 2A
>>  option remote-subvolume brick
>> end-volume
>>
>> volume remote2
>>  type protocol/client
>>  option transport-type tcp/client
>>  option remote-host 2B
>>  option remote-subvolume brick
>> end-volume
>>
>> volume mirror0
>>  type cluster/afr
>>  subvolumes remote1 remote2
>> end-volume
>>
>> I do see this message during the load, but I understand that it's harmless:
>>
>> 2008-06-18 13:00:21 D [xlator.c:115:xlator_set_type] xlator:
>>  attempt to load file
>> /usr/local/lib/glusterfs/1.3.9/xlator/features/posix-locks.so
>>
>> 2008-06-18 13:00:21 D [xlator.c:145:xlator_set_type] xlator: dlsym(notify)
>>  on /usr/local/lib/glusterfs/1.3.9/xlator/features/posix-locks.so:
>> undefined symbol: notify -- neglecting
>>
>> Thanks in advance,
>> Dale Johnson
>>
>>
>> _______________________________________________
>> Gluster-devel mailing list
>> address@hidden
>> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>



-- 
Dale Johnson (address@hidden)