Thursday, August 11, 2022

[fedora-arm] Re: RPi (or something similar) cluster and GlusterFS - feedback?

On Thu, 11 Aug 2022, Philip Rhoades wrote:

> Has anyone here set up a cluster something like this? Have people any
> suggestions about specific Fedora web pages / docs to look at?

It's pretty simple to set up iSCSI. For performance, you might want to
dedicate a 1Gbit ethernet LAN, but it shares the LAN just fine.
It works just a remote SATA, but potentially longer cable.
Also, I run LVM on an iSCSI server, which exports logical volumes as
iSCSI drives.

The rPi4 is fast enough. A caveat for me is I like ECC ram for long
running stuff. But you could use btrfs, or run the data integrity layer
in device mapper.

I use command line to allocate volumes, export them as iSCSI, mount them
on client. VPS shops have webapps to allocate volumes, map them to a
VPS, etc - and I suspect there is some sort of "dashboard" in Fedora,
but I don't know about it.

I also highly recommend DRBD - Data Replicating Block Device. This will
send writes to a volume over the network to another server, so that the
remote server is a mirror. It works between sites for instant offsite backup
if the write volume is low enough. On a local LAN, you can live migrate
a VM between servers using only DRBR. iSCSI also works for that, of

> I have been tracking CEPH for some time and while it looks interesting, it
> seems overkill for what I am thinking of doing and more technically difficult
> to support / debug etc. I am thinking of starting the exercise with 4 RPis
> each with an 8Tb SATA drive.

I found CEPH overkill for my purposes also. I call what we want,
"Poor Man's Cluster".
arm mailing list --
To unsubscribe send an email to
Fedora Code of Conduct:
List Guidelines:
List Archives:
Do not reply to spam, report it:

No comments:

Post a Comment