I was thinking about storing indices files on several disk partitions.
Why? Because some indices will be huge and "low risk" (indexing logs) and
others won't be that big but will be sensitive.
That would allow me to set different alert levels on partition usage, and
maybe make things safer for index files?
So my questions:
Is storing indices on different partitions (on the same physical disk)
doable/a good idea?
Or could it trigger problems with cluster state?
Maybe there's no point at all?
I was thinking about storing indices files on several disk
partitions. Why? Because some indices will be huge and "low risk"
(indexing logs) and others won't be that big but will be sensitive.
That would allow me to set different alert levels on partition
usage, and maybe make things safer for index files?
So my questions:
Is storing indices on different partitions (on the same physical disk)
doable/a good idea?
Or could it trigger problems with cluster state?
Maybe there's no point at all?
It's no problem operationally, although there will probably be little
value in it. If it makes your life easier monitoring disk usage, go
for it. It won't affect cluster state and probably won't have any
effect on performance.
That said, operating systems and hdd/ssds are complicated things
nowadays and it could dramatically help or hinder either way. It's
always best to profile it with a large amount of your data.
How would you achieve it? Symlinks?
You can use symlinks, but check out the multiple data locations
example here:
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.