[Gluster-devel] cache doesn t seem to be taken into account

[Sebastien COUPPEY]

On Thu, Jan 10, 2008 at 03:17:50PM +0100, Sascha Ottolski wrote:
> Am Donnerstag 10 Januar 2008 15:02:15 schrieb Sebastien COUPPEY:
> > Thanks,
> > and the compilation on the debian system run smoothly so I performed
> > an other test.
> >
> > The load balancing and the AFR is working fine. Howere, during the
> > copy a a huge amount of files the client crash with the following
> > error :
> >
> > ---------
> > got signal (11), printing backtrace
> > ---------
> > [0xffffe420]
> > /usr/local/lib/glusterfs/1.3.8/xlator/cluster/afr.so(afr_setdents+0x1
> >ca)[0xb7543bda]
> > /usr/local/lib/glusterfs/1.3.8/xlator/cluster/unify.so(unify_sh_ns_ge
> >tdents_cbk+0x17b)[0xb7538c2b]
> > /usr/local/lib/glusterfs/1.3.8/xlator/cluster/afr.so(afr_getdents_cbk
> >+0x208)[0xb7544418]
> > /usr/local/lib/glusterfs/1.3.8/xlator/protocol/client.so[0xb7f0fb2c]
> > /usr/local/lib/glusterfs/1.3.8/xlator/protocol/client.so(notify+0xb50
> >)[0xb7f109c0]
> > /usr/local/lib/libglusterfs.so.0(transport_notify+0x37)[0xb7efeba7]
> > /usr/local/lib/libglusterfs.so.0(sys_epoll_iteration+0xd7)[0xb7eff8f7
> >] /usr/local/lib/libglusterfs.so.0(poll_iteration+0x98)[0xb7efecf8]
> > [glusterfs](main+0x4a6)[0x804a906]
> > /lib/i686/cmov/libc.so.6(__libc_start_main+0xe0)[0xb7d77450]
> > [glusterfs][0x804a3f1]
> > ---------
> >
> >
> > 1.3.8 is probably under huge development ... So I guess I will have
> > to find a work around if we use 1.3.7 in production.
> >
> > Thanks a lot matthias
> 
> just a guess, I think I've seen a similar error with a patchlevel > 628, 
> so maybe your checkout is a bit too recent? At least for me, patch-628 
> seems to run quite stable.

patch634, the tla checkout of this morning.

I recreated the node, and realised that first I missed the
"user_xattr" in the mount options.

Now with the 1.3.7 the AFR and and UNIFY I can see the request and
exchanges from the both nodes.


I realised that the size of the folders is quite different :

 source (12 Mo=3051 files) 
     |
     | copy to GLFS cluster
    \ /
  GLFS folder (24Mo=3051 files)
     |
     | copy to a client
    \ / 

 ext3 Fs (12 Mo = 3051files)


It looks strange to me that files on the GLFS are taking 2 times the size
is it normal ?