search

fengshao0907 / weed-fs

Automatically exported from code.google.com/p/weed-fs
0 stars 0 forks source link

replica placement improvement #67

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
When data center or racks has un-even number of volume servers, the replication 
"110" may fail due to it randomly starting from one of the volume servers. 

The work around is to use preferred data center. However, this needs manual 
operation.

For example, here is a user's case:

I start 6 volumes named "volume1" to "volume6" respectively. Set max=100. And 
the topology is:
    datacenter1:
        rack1:volume1, volume2
        rack2:volume3
    datacenter2:
        rack1:volume4, volume5
        rack2:volume6
    And then I want to upload some files with replication=111, but it dosen't work and show me error: Not enough data node found. I do have 2 different datacenter and racks, why it can't work?

Original issue reported on code.google.com by chris...@gmail.com on 10 Apr 2014 at 9:24

GoogleCodeExporter commented 9 years ago 
v0.53 when growing volumes, additional preferred "rack" and "dataNode" 
parameters are also provided, works together with existing "dataCenter" 
parameter.

Original comment by chris...@gmail.com on 13 Apr 2014 at 8:34

GoogleCodeExporter commented 9 years ago 
v0.53 fixed this bug.

Original comment by chris...@gmail.com on 13 Apr 2014 at 10:23