cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jeff Jirsa"<jji...@apache.org>
Subject Re: jbod disk usage unequal
Date Mon, 03 Jul 2017 18:01:54 GMT


On 2017-06-29 06:55 (-0700), Micha <micha-1@fantasymail.de> wrote: 
> Hi,
> 
> I use a jbod setup (2 * 1TB) and the distribution is a little bit
> unequal on my three nodes:
> 270MB and 540MB
> 150 and 580
> 290 and 500
> 
> SStable size varies between 2GB and 130GB.
> 

You're switching between MB and GB, are these all in GB? 

> Is is possible to move sstables from one disk to another to balance the
> disk usage?

Is there a reason you feel it's required, other than being bothered by the fact that they're
not equal? 

> Otherwise is a raid-0 setup the only option for a balanced disk usage?

Probably for most versions of cassandra (newer versions will attempt to choose JBOD disk based
on token ranges, which is designed to protect against some other edge case bugs where you
lose one disk and have to rebuild the whole node anyway). The real question is "why do you
care, really?" - it'll spill over to the other drive when it needs space, and if it's not
near capacity, it doesn't REALLY matter how full it is, right?



---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@cassandra.apache.org
For additional commands, e-mail: user-help@cassandra.apache.org


Mime
View raw message