Sheng Yang

I am trying to understand why we need to be concern about the internal traffic originated from the cluster. Longhorn UI is exposed as a service by default, which is...

The goal (ranked): 1. Able to provision bare-metal servers on the same L2 network with arbitrary ISO or cloud image. 2. Inventory management to know what's in use by who...

Why do we need `iostat` here? FIO already has the categorized results. And it's not a good idea to bind mount a host directory just to copy the files out,...

I see. A few comments here: 1. Is there a way for us to figure out the block device without specifying it manually? 2. Does it work if we pass...

@jcambronero We haven't scheduled this feature yet. What's your use case? Currently, we're recommending using LVM to aggregate disks on one node to help.

In that case, yes you will need this feature. We are in the development cycle of v1.1.0 release, targeting October. Once we're done with v1.1.0, we can take a look...

@PhilParisot we've raised the priority of this issue. We will start working on it in the near future.

Also, as an alternative, we can consider LVM in striping mode. But need to know what we should do to deal with the disk failure.

Can you try https://github.com/yasker/kbench ? In your fio job, you're testing bandwidth with 4k block, which is mostly used for IOPS test (since the block size is small). Also with...

@liyimeng I think your latest result looks valid. There is performance overhead on using Longhorn. Though Longhorn should still be better than most other software-defined distributed storage solutions out there...