[elrepo] BTRFS for RHEL 8

Pat Riehecky riehecky at fnal.gov
Wed Jan 22 09:47:14 EST 2020



On 1/21/20 5:49 PM, Konstantin Olchanski wrote:
> On Tue, Jan 21, 2020 at 09:13:33PM +0000, Joseph Christopher Sible wrote:
>> Red Hat removed all traces of BTRFS from RHEL 8. However, I did some searching online and found
>> evidence that ELRepo has packages that bring it back. https://urldefense.proofpoint.com/v2/url?u=https-3A__unix.stackexchange.com_a_545089&d=DwICAg&c=gRgGjJ3BkIsb5y6s49QqsA&r=OAMtP0DWou0nlXG7Kmxo2enjXJfwb1DXS9fwcaESuTE&m=XdmEgeMDA5EQ7iXGlhVXFjosXcp1CEbRWPYCjw5gJ6w&s=DpNVI-u9x37wvA94j1EzkFn-kUuefnVrOAGB6E9w9jI&e=  and
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__centosfaq.org_centos_missing-2Dpackages-2Din-2Dcentos8-2Dmirrors-2Dor-2Ddo-2Di-2Dmiss-2Dsomething_-23comment-2D128436&d=DwICAg&c=gRgGjJ3BkIsb5y6s49QqsA&r=OAMtP0DWou0nlXG7Kmxo2enjXJfwb1DXS9fwcaESuTE&m=XdmEgeMDA5EQ7iXGlhVXFjosXcp1CEbRWPYCjw5gJ6w&s=Uz5r5q4VfWSqlBii7o-9DzvJoYNXNUQq3W1A_Lqm4wk&e=
>> both seem to say that kmod-btrfs and btrfs-progs are available in ELRepo 8 Testing. However, I don't
>> see them there. What am I missing? Where are they?
> FWIW, a few years ago I evaluated btrfs as shipped in centos-7.1 (or 7.2 or 7.3)
> for replacement of mdadm raid6+xfs for small storage arrays (up to 8 disks).
>
> In my evaluation, many btrfs things worked very well (better than zfs), but
> I immediately ran into a number of problems that are show stoppers and
> (the best I can tell) are still not fixed:
>
> - failure of one disk resulted in filling the syslog with an infinite
>    stream of errors (vs mdadm raid and zfs faulting-out the bad disk)
> - systemd refused to boot in the presence of a degraded btrfs filesystem
> - raid5/raid6 are still labeled "unstable, write hole exists", fix proposed
>    in 2017 (today is 2020), see https://urldefense.proofpoint.com/v2/url?u=https-3A__btrfs.wiki.kernel.org_index.php_RAID56&d=DwICAg&c=gRgGjJ3BkIsb5y6s49QqsA&r=OAMtP0DWou0nlXG7Kmxo2enjXJfwb1DXS9fwcaESuTE&m=XdmEgeMDA5EQ7iXGlhVXFjosXcp1CEbRWPYCjw5gJ6w&s=HU-3oQdNUzdqD7y9d3AWvhFuTUFi2fIgzDrCFnx0ivw&e=
>
> Since then, I switched from mdadm+xfs to ZFS (mirrored home directories and
> raid6 data arrays). It works well enough.
>
> When btrfs has stable raid5/raid6 I will look at it again.
>
> I would think Red Hat did a similar evaluation with similar results
> leading to their decision to remove btrfs completely.
>
> P.S
>
> If the OP succeeds in installing btrfs, I would like to hear from them
> about the latest status of booting in the presence of degraded btrfs
> filesystems (i.e. one disk missing from mirror or raid5/raid6). Last time
> I tried, booting stalled in systemd waiting for the missing (failed) disk
> to come online. All suggested workarounds (about 3 or 4 of them) did not work.
>

I can't comment on the btrfs bits exactly, but has anyone played around 
with stratis?

https://stratis-storage.github.io/
https://www.redhat.com/en/about/videos/rhel-8-beta-managing-storage-stratis

To my eye, it seems to be where RH is targeting the btrfs workloads...

Pat

-- 
Pat Riehecky

Fermi National Accelerator Laboratory
www.fnal.gov
www.scientificlinux.org



More information about the elrepo mailing list