One time I launched a _dump on all the indexes available on my elasticsearch cluster, the node suffered of an overloading.
The node, in normal working, has a system load about 1. When the _dump was launched, system loads has grown up to 22.
I think _dump on several indexes tries to dump shards as many as possible.
I think doing the _dump one index a time is a better way to do.
However, what happens if you have many shards for an index ?
Is it possible to manage a sort of kindness inside plugin ? (Max dumping shards at a time, ...)
One time I launched a _dump on all the indexes available on my elasticsearch cluster, the node suffered of an overloading. The node, in normal working, has a system load about 1. When the _dump was launched, system loads has grown up to 22. I think _dump on several indexes tries to dump shards as many as possible.
I think doing the _dump one index a time is a better way to do.
However, what happens if you have many shards for an index ?
Is it possible to manage a sort of kindness inside plugin ? (Max dumping shards at a time, ...)