Re: [Moosefs-users] Replication mechanism/speed

[Reinis R. <r...@ro...>]

> Wouldn't you have more chunks, then you would have sufficient redundancy 
> to
> cover a further failure that the 10 days wouldn't be important ?

Of course, but that was not the question - I would just like to know the 
mechanism/algorithm behind without diving in the source code.
Own tests showed that just by adding few more chunkservers the replication 
speed goes up so it partly answers my own innitial mail (just if possible 
wanted some detailed clarification from devs).


> 1+ million files created must be extreme usage.

Yes, we are migrating from a GlusterFS (so far more just in test scenario 
though) solution - with 200+ million files.

rr