Red Hat Enterprise Linux 8 Beta Released With Stratis, Yum 4, Application Streams

[cjcox]

...
Instead of doing more wheel reinventing with Stratis, Red Hat could have helped wrap up loose ends with btrfs and gotten it ready for production use. That would give something which is much more flexible than Stratis.

Certainly would have been nice to see. But Red Hat doesn't play well with open source it doesn't "own". it was worse for Red Hat before they acquired Sistina and XFS. Red Hat's view of open source probably isn't like most people's view of open source. Red Hat has specific motives behind what they do that are often in conflict with open source as a whole. Which probably is why IBM acquired them. Red(dington) Hat.

[SuperIce97]

A surprise to be sure, but a welcome one. Glad to see IBM speeding up the release in just a few weeks.

This has nothing to do with IBM. IBM has not yet acquired Red Hat (only announced their intention to) and the process will take ~6-9 months to finalize. Until then, Red Hat is not allowed to change their behavior or business plans as a response to the acquisition announcement. I will also say that the release date for RHEL 8 beta was on their internal (confidential) calendar for many months now, long before IBM came into the picture. It just wasn't public (in case anything important changed and for legal reasons). There have been internal betas that were being tested within Red Hat for some time as well (before the IBM announcement).

[garegin]

I sure hope IBM smacks some sense into them in supporting Btrfs and also paying more devs in finishing Wayland.

[HadrienG]

Crossing fingers. Maybe my sysadmin colleagues will manage to do the CentOS 8 upgrade faster than the (still ongoing) CentOS 7 one, so that for a brief short period of time I am able to reliably profile code with perf directly on the compute nodes instead of spending ages replicating the whole setup on my desktop.

And then the kernel will become obsolete with respect to current-gen hardware again, as customary on Red Hat.

[birdie]

This is going to be a fabulous release. My only gripe is that it's based on F28 as opposed to just released F29.

[cen1]

This has nothing to do with IBM. IBM has not yet acquired Red Hat (only announced their intention to) and the process will take ~6-9 months to finalize. Until then, Red Hat is not allowed to change their behavior or business plans as a response to the acquisition announcement. I will also say that the release date for RHEL 8 beta was on their internal (confidential) calendar for many months now, long before IBM came into the picture. It just wasn't public (in case anything important changed and for legal reasons). There have been internal betas that were being tested within Red Hat for some time as well (before the IBM announcement).

bro.. it was a joke

[jpg44]

Agree. I personally use ZFS on Linux, hopefully one day btrfs will be ready as well.

ZFS is by far the best thing going on Linux. Admittedly, btrfs is still at the point you wonder about trusting it. Perhaps with Stratis Red Hat may develop a more of a COW snapshot and versioning aware filesystem on top of the devmapper, using the dev mapper for low level block management, who knows, perhaps they are aiming to evolve into that. Its not that Stratis is terrible, might be the best in some environments, but not really the state of the art at this point. It depends on what your requirements are.

Linux for a while has been behind Solaris and BSDs on the filesystems, so while they got ZFS, the Linux land didnt have anything that quite matches that (unless you ignore the license issues and use ZFS).

[thunderbird32]

What's the current consensus on ZFS licensing? I was under the impression that it's still a bit of an unresolved issue. I would assume this would make Red Hat a bit gun-shy about using it. That said, I can't see why they didn't look to BTFS instead (not that XFS has ever given me any issues).

[Candy]

I'm almost sure it will have dnf 4 - the same version as in F29.

Hopefully not. The DNF Version that got delivered with Fedora 29 has a huge regression and behaves differently to DNF that was delivered all previous versions before Fedora 29.

[Anvil]

Stratis is a big step back from the next generation filesystem we could have with forward looking designs, either btrfs or moving toward Cephfs has a general purpose filesystem even. Stratis is a 1980s filesystem on top of a 1980s block layer. When there is need for a versioning filesystem and git-like features to track changes being made to a filesystem, it seems that running xfs over lvm is quite a primitive and clumsy solution. Every minute change to a file requires a complete fork of the entire XFS filesystem. It sounds like this would be unwieldy if one just wants to track small changes to individual files or turn on versioning only for particular files.

It seems like Stratis is a nothingburger, anyone can just throw an XFS filesystem on top of an LVM. If your looking for per file versioning, its really going to be awkward. It would be much more natural and considered to be a better design to integrate these features such as versioning and snapshots into a filesystem that can natively support COW writes and natively track file versions, rather than rely on a block layer.

Instead of doing more wheel reinventing with Stratis, Red Hat could have helped wrap up loose ends with btrfs and gotten it ready for production use. That would give something which is much more flexible than Stratis.

stratis is NOT a Filesystem. where does it say that it is ?