block size problem ?

Hi peter, 
Now my redundant setup is working on production servers. I have intelligent
reconstruction from both
side. Perfs are pretty good. Without any tweaking,  36 MB/s during
reconstruction over GbE link, and bonnie between 12-20 MB/s Writes / 20-50 MB/s
 reads. It's a little bit less for writes thant with direct access to disk, and
slightly more for reads.

have you recieved my 2.6.12 patches two weeks ago ?

I'm writing a little howto for my new heartbeat scripts. for the moment it is in
french, but i can translate it. i'll sent you the scripts and doc soon.

There is just a little problem remaining : enbd block size is set at 4096 both
size. Disk is formatted in XFS, with blocksize 4096 and I have this : Is it
harmful ?

enbd-server 21903: <#1405> do_srv_write request for nonblock (4096) offset at
0x13004f1a00 (81609562624) seqno 2 len 65536
enbd-server 21903: <#1408> do_srv_write are you sure you really wanted blksize
4096? The kernel seems to issue requests aligned at 512
enbd-server 21900: <#1405> do_srv_write request for nonblock (4096) offset at
0x1300501a00 (81609628160) seqno 3 len 65536
enbd-server 21900: <#1408> do_srv_write are you sure you really wanted blksize
4096? The kernel seems to issue requests aligned at 512
enbd-server 21901: <#1405> do_srv_write request for nonblock (4096) offset at
0x1300511a00 (81609693696) seqno 4 len 65536
enbd-server 21901: <#1408> do_srv_write are you sure you really wanted blksize
4096? The kernel seems to issue requests aligned at 512
enbd-server 21902: <#1405> do_srv_write request for nonblock (4096) offset at
0x1300521a00 (81609759232) seqno 5 len 65536
enbd-server 21902: <#1408> do_srv_write are you sure you really wanted blksize
4096? The kernel seems to issue requests aligned at 512
enbd-server 21902: <#1414> do_srv_write request for nonblock (4096) length at 0
(0) seqno 8 len 512


Denis

----------------------------------------------------------------
      Lycée Diderot Paris  - Ce message est envoyé avec IMP -