On 01/07/2015 12:48 PM, Raghavendra Bhat wrote:
Hi,
As per the design dicussion it was mentioned that, there will be one
BitD running per node which will take care of all the bricks of all
the volumes running on that node. But, here once thing that becomes
important is doing graph changes for the BitD process upon
enabling/disabling of bit-rot functionality for the volumes. With more
and more graph changes, there is more chance of BitD running out of
memory (as of now the older graphs in glusterfs are not cleaned up).
Both NFS and SHD processes have same problem, but upon graph switch the
daemons are restarted. Is there any reason this approach is not taken?
Pranith
So for now it will be better to have one BitD per volume per node. In
this case, there will not be graph changes in BitD. It will be started
for a volume upon enabling bit-rot functionality for that volume and
will be brought down when bit-rot is disabled for a volume.
Regards,
Raghavendra Bhat
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxxx
http://www.gluster.org/mailman/listinfo/gluster-devel
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxxx
http://www.gluster.org/mailman/listinfo/gluster-devel
