[Gluster-devel] Fwd: glusterfs write problem

Anand Avati avati at zresearch.com
Wed Apr 9 05:09:55 UTC 2008


nicolas,
 an aggregate-size of 10MB (in write-behind) is just too high to be used on
the client size. please unset it and try.

avati

2008/4/9, nicolas prochazka <prochazka.nicolas at gmail.com>:
>
> Hi,
>
>   I'm working with file of 10 Go size.
>   When I read file all is ok and seems to be worked fine with glusterfs
>   If I read and write one file in the same time, all is bad :
>   - gluster client and server takes a lot of resources ( 30 %- 60 % of
> cpu)
>   - write is very very slowly and does not work  , read also seems to
>   be cycling
>
>   I'm trying with two configuration
>
>   Computer 1 : client
>   Computer 2 : server
>
>   computer1 ---> Read big file and write in local   <---> computer2   :
> works fine
>   computer1 ---> Read and Write big file  <---> computer2  : not work
>   computer2 become client also so i  mount gluster in local :  Read and
>   Write big file : not work. (gluterfs and glusterfsd take a lot of
>   ressource)
>
>   I'm trying differents client / server configuration without success
>
>   Any idea ?
>   Regards,
>   Nicolas Prochazka.
>
>
>
>   things to know :
>   - glusterfs 1.3.8pre5
>   - fuse : fuse-2.7.2glfs9
>
>   ------------------
>   Computer 2 : Server configuration
>   ----------------------------------------------------
>   volume brick1
>   type storage/posix
>   option directory /mnt/disks/export
>   end-volume
>
>
>   volume brick
>   type performance/io-threads
>   option thread-count 8
>   option cache-size 1000MB
>   subvolumes brick1
>   end-volume
>
>
>   volume readahead-brick
>   type performance/read-ahead
>   option page-size 2M
>   option page-count 128
>   subvolumes brick
>   end-volume
>
>
>
>   volume server
>   option window-size 2097152
>   type protocol/server
>   subvolumes readahead-brick
>   option transport-type tcp/server     # For TCP/IP transport
>   option client-volume-filename /etc/glusterfs/glusterfs-client.vol
>   option auth.ip.brick.allow *
>   end-volume
>
>
>
>   ----------------------------------------
>   computer 1 ------ client
>   -----------------------------------------
>   volume client1
>   option window-size 2097152
>   type protocol/client
>   option transport-type tcp/client
>   option remote-host 10.98.98.1
>   option remote-subvolume brick
>   end-volume
>
>   volume readahead
>   type performance/read-ahead
>   option page-size 2MB
>   option page-count 64
>   subvolumes client1
>   end-volume
>
>   volume iothreads
>   type performance/io-threads
>   option thread-count 32
>   subvolumes readahead
>   end-volume
>
>   volume io-cache
>   type performance/io-cache
>   option cache-size 1000MB             # default is 32MB
>   option page-size 1MB               #128KB is default option
>   option force-revalidate-timeout 100  # default is 1
>   subvolumes iothreads
>   end-volume
>
>   volume writebehind
>    type performance/write-behind
>    option aggregate-size 10MB # default is 0bytes
>    option flush-behind on      # default is 'off'
>    subvolumes io-cache
>   end-volume
>
>
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at nongnu.org
> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>



-- 
If I traveled to the end of the rainbow
As Dame Fortune did intend,
Murphy would be there to tell me
The pot's at the other end.



More information about the Gluster-devel mailing list