Hello,
i am new in gluster. before i tried moosefs with goal=1 for my data. now i want to give glusterfs 3.4 a try. my testsystem includes 2 clients and 4 server / bricks. i startet with 2 bricks and vol1 as a distributed filesystem. i noticed quick that glusterfs will fill first eine brick and after i think a percent of free disk on brick1, the second brick with data. ok that was not my aim but describes the distributed filesystem.
after that i reinstall and install vol1 with typ striped (1*2=2). that was fine, the data was written on both bricks simultaneously and i get a good performance. after that i added 2 more bricks and have 2*2=4 with type striped. my first test show, that only brick 3 and 4 will filled with data simultaneously and not all 4 together. ok i can understand this construction, but when will glusterfs use brick1 and brick2? is there any case in this construction where glusterfs will writes on all 4 at the same time? and am i right that i have to change the count of replica to 1*4=4 to write on all bricks simultaneously?
my am is to get the most performance out of the cluster. security will lay above on the raid6 of each brick.
thx for explaining and describing
sunny
_______________________________________________ Gluster-users mailing list [email protected] http://supercolony.gluster.org/mailman/listinfo/gluster-users
