I doubt mv will run instantly because copy is across two different
On Sun, 24 Jun 2018 at 9:26 PM, Nitan Kainth <nitankai...@gmail.com> wrote:
> To be safe you could follow below prices on each node one at a time:
> Stop Cassandra
> Move sstable— mv will do it instantly
> Start Cassandra
> If you do it online and a read trust comes for sane data that is being
> moved will fail.
> On Jun 23, 2018, at 11:23 PM, onmstester onmstester <onmstes...@zoho.com>
> I'm using two directories on different disks as cassandra data storage,
> the small disk is 90% full and the bigger diskis 30% full (the bigger one
> was added later that we find out we need more storage!!),
> so i want to move all data to the big disk, one way is to stop my
> application and copy all sstables from small disk to big one, but it would
> take some hours and not acceptable due to QoS.
> I thought maybe i could copy the big sstables (the one that won't be
> compact in weeks) to the big disk (near casssandra data but not right
> there) while cassandra and my app are still running
> , then stop cassandra and my app, move big file to exact directory of
> cassandra data on big disk (would take a few seconds) and then move
> remained small sstables from small disk to big one.
> Are there all of sstables related file immutable and (data, index,
> summary, ...) would only be changed by compactions? Any better workaround
> for this scenario would be appriciated?
> Thanks in Advance
> Sent using Zoho Mail <https://www.zoho.com/mail/>