[Gluster-devel] gcc version dependency?

[rhubbell]

On Sun, 2008-11-09 at 00:18 +0530, Anand Avati wrote:
> Do you know if solaris has __BYTE_ORDER defined in /usr/include/* ?
> Can you give us access to this particular machine to inspect a few
> macros? It seems to be an endian-ness related issue.
> 

BTW, I was able to telnet to the server port from the client.
I guess you must mean it's not a network byte-order problem?

> avati
> 
> 2008/11/9 rhubbell <rhubbell at ihubbell.com>
>         On Sat, 2008-11-08 at 17:10 +0530, Anand Avati wrote:
>         >
>         >
>         >On Fri, 2008-11-07 at 10:26 -0800, rhubbell wrote:
>         >> > Hello all,
>         >> >
>         >> > Is there a version dependency on gcc and gluster?
>         >> > Is there currently a preference by devs? gcc3 or 4?
>         >> >
>         >> >
>         >>
>         >> I ended up building with gcc4 and I got it built on
>         Solaris10.
>         >> I also built it on redhat linux from the same tar ball,
>         1.4pre7
>         >>
>         >> I started with a simple setup and I am unable to get the
>         linux box to
>         >> mount the solaris gfs.
>         >>
>         >> I installed the fuse-gfs and built that and loaded the
>         kernel module
>         >> fine on the linux client.
>         >>
>         >> Error on client:
>         >>
>         >> client: SETVOLUME on remote-host failed: ret=-2
>         reply=@0x8fcd270
>         >>
>         >> ls on the mount point produces:
>         >>
>         >> Transport endpoint is not connected
>         >>
>         >>
>         >>
>         
>         
>         >> server errors (not exact):
>         >>
>         >> server: socket header validation failed
>         >> socket.c:492  socket header has incorrect size1=3120562176
>         >>
>         
>         >
>         > you are probably running 1.3.x against 1.4.x
>         >
>         > avati
>         >
>         
>         Not sure why you reached that conclusion as I specifically
>         pointed out
>         that I built from the same tar ball on solaris and linux.
>         
>         I have to guess that something in the error is familiar.
>         
>         I confirmed that I'm running the same version on client and
>         server.
>         
>         
>         Here's some additional data, if you have better ideas to
>         troubleshoot or
>         other data that would help to narrow this down I'm all ears.
>         (^:
>         
>         
>         
>         Server (running solaris10 - from the log file):
>         ---------------------------
>         
>         2008-11-08 13:27:58 E
>         [socket.c:492:__socket_proto_validate_header]
>         server: socket header has incorrect size1=3120562176
>         2008-11-08 13:27:58 E
>         [socket.c:588:socket_proto_state_machine] server:
>         socket header validation failed
>         2008-11-08 13:27:58 C
>         [server-protocol.c:7176:server_protocol_cleanup]
>         server: connection private (null) for transport 7dad8
>         2008-11-08 13:27:58 D [socket.c:1292:fini] server: transport
>         7dad8
>         destroyed
>         
>