Hello.
We have path.data with 2 paths in elasticsearch.yml:
path.data: /path/data1,/path/data2
Initially /path/data1 and /path/data2 were on the different block devices (that's why there are 2 of it).
But after reconfiguring disk subsystem on the physical node, they are now on the same block device.
So, the question: are there any performance issues with having multiple path.data on the same block device? Should we merge it somehow, and if we should, what is the simplest way to do it properly?
Hi Sergey,
we don't check whether multiple paths map to the same physical device. So we treat this as if it were two separate devices which could lead to an increase in random disk I/O and this will likely hurt performance.
So if you want to avoid that you need to change the setting and reload the data. You can use snapshot / restore for that. If Elasticsearch is just a secondary data store you can also just import the data again into Elasticsearch (whatever works better for you).
Daniel