Re: [Moosefs-users] mfs-master[4166]: CS(10.10.10.10) packet too long (226064141/50000000)

[Fabien G. <fab...@gm...>]

Hi Michal and moosefs-users@,

2010/6/23 Michał Borychowski <mic...@ge...>

> Probably important is the difference in the amount of chunks per one
> chunkserver. We have about 800,000 chunks per chunkserver (60 million chunks
> on 75 machines).
>

Thanks for you quick answer. Yes you're right, that's what I thought too :
4.7M chunks per chunkserver is far bigger than 800K

Just a question : How much RAM do you use on the master, and on slaves ? In
our case (11'800'000 chunks on 5 chunkservers) :
* 'mfsmaster' process on master : 4.9 GB (64 bits recompilation required :
the 32 bits version of mfsmaster crashed without a message when it came to 4
GB)
* 'mfschunkserver' process on chunkservers : 580 MB


 How many files do you have? What is the average size of a file? What goal
> do you have set?
>

Our current test cluster is used for backups storage. We have lots of files
from all sizes (rsync of /etc, /home/, ... files on several servers), and
also a lot of big archive files (several GB each). We currently have 14.7
millions inodes used.

Maybe later, we'd like to use it for webhosting. But since LOCK is not
supported, it's not yet possible.


Fabien