Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Rocky Linux 9.0

rockylinux.org

Post reply on HN
Print view View original

101–110 of 178 posts

Re: Rocky Linux 9.0

#101

Post by tokamak »

jillesvangurp wrote: post #98

I find it interesting that there's a need for Rocky to exist at all. IBM basically killed Centos, which was the unofficial Red Hat release that came into existence when Red Hat stopped providing free versions of Red Hat which Red Hat then grand fathered in as a thing they actively supported. That whole process was just weird. I did not encounter a single project that actually used Red Hat after that happened: world +…

RedHat benefit is support. You can sue them if something goes wrong ;) Good engineering is using on local what is used on servers. RHEL definitely is about servers foremost. There are many industries where RHEL is a must. Amazon Linux is derivative of RHEL. Good point about containers.

Re: Rocky Linux 9.0

#102

Post by josephcsible »

bonzini wrote: post #99
josephcsible wrote: post #97

Earlier quoted context omitted.

This isn't my experience at all. I've personally had three paid support cases attached to Bugzilla issues. Two of them were for bugs that had one-line fixes that were already committed upstream, along with test cases, and that I just needed Red Hat to backport. They sat for about 6 months before I got any response other than the automated monthly "our engineers are working on the Bugzilla." The third was for a proble…

BZ number for the third? Also what components?

The components for the first two were NSS (Network Security Services, not Name Service Switch) and GCC, and the third was Kerberos. I don't have the BZ number handy right now.

Re: Rocky Linux 9.0

#103

Post by liamnal »

5e92cb50239222b wrote: post #17
bubblethink wrote: post #13

Earlier quoted context omitted.

They are both new. So, this may be a one time thing due to tooling changes or whatever. Too early to jump to conclusions.

It's been close to 1.5 years. I've been using both since the beginning and can see pretty well how quickly both systems pick up updates. But let's just look at release delays (in days since the official RHEL is shipped): ver Alm Rocky 8.4 8 34 8.5 3 6 8.6 2 6 9.0 9 58 Not seeing any patterns here? One of them is being done by a team that's been shipping another Linux distribution for a decade and has the whole proces…

Rocky made it clear they have a new build system which lead to their delay. I think it's called peridot. I would imagine that it's not easy to start fresh in two places at once. Alma has an advantage with a process from CloudLinux already in place and now they've rebranded that build system as Alma and so on. Credit to Alma for keeping up and doing their thing; hopefully they'll be fully decoupled from CL in the future.

CentOS had historically fell behind in release times, but we all of a sudden want to paint others in a bad light for being behind. Did everyone forget about CentOS 7 having average of 30 days delay behind each point release? 7.4 being the most at 43 days. What about CentOS 6.0, with 242 days?

As for their behavior, it takes two to tango. I usually like to thank carlwshill and "conan_kudo" (who should pick a better name since he thinks he's from the anime his picture is from) for being true stars in the open source community and really bringing out not only the best in others, but the best in themselves day in and day out. I'm not sure how anyone can put up with them, regardless of which community you're in (CentOS/Fedora EPEL/others), but who am I to judge, I'm just a user.

Re: Rocky Linux 9.0

#104

Post by chasil »

tokamak wrote: post #101
jillesvangurp wrote: post #98

I find it interesting that there's a need for Rocky to exist at all. IBM basically killed Centos, which was the unofficial Red Hat release that came into existence when Red Hat stopped providing free versions of Red Hat which Red Hat then grand fathered in as a thing they actively supported. That whole process was just weird. I did not encounter a single project that actually used Red Hat after that happened: world +…

RedHat benefit is support. You can sue them if something goes wrong ;) Good engineering is using on local what is used on servers. RHEL definitely is about servers foremost. There are many industries where RHEL is a must. Amazon Linux is derivative of RHEL. Good point about containers.

I'm not sure if Suse Liberty Linux ISOs are fully available yet, but that is yet another clone offering support.

https://www.suse.com/c/suse-liberty-linux/

Oracle Linux 9 also dropped a week ago, and their support is available if increasingly expensive.

The problem with both of these is app support certifications. Microsoft SQL Server (for example) is solely supported on RHEL, although it's obvious that they build on CentOS.

Re: Rocky Linux 9.0

#105

Post by bonzini »

linuxftw wrote: post #76
bogwog wrote: post #27

Earlier quoted context omitted.

You're forgetting to mention the part where Rocky Linux is not just a rebrand of RHEL, it's a revival of CentOS, which IBM killed off in what's effectively a bait-and-switch, forcing customers to go through the painful/expensive migration process to another distro, or the less-painful but still expensive migration process to RHEL. Rocky Linux is a shining example of both a free market and the open source community wo…

Sadly, the decision to kill CentOS was entirely Red Hat's. The self-appointed 'community managers' decided that the community didn't actually want CentOS, they wanted a free version of RHEL-Beta called "CentOS Stream."

Nobody said it was a community choice. Red Hat's own layered products (OKD, RDO, oVirt) needed to be based on a distribution that leads RHEL rather than one that trails it, and that's why CentOS Stream was born.

Dropping CentOS Linux was a completely different thing and one should also acknowledge that there are two very different parts of the CentOS community.

Those that simply needed a free RHEL, didn't have any benefit from CentOS Stream. However, their usecase is filled by Alma/Rocky.

Downstream CentOS distributions however only got benefits from CentOS Stream. There are many private ones, for example Facebook runs on a CentOS derivative, but the most prominent example is Alma itself, which existed (IIRC with another name) even when CentOS Linux existed.

And to be honest, only the latter are really part of the community. Downloading an ISO doesn't make you part of the community. I myself used CentOS Linux on a small EC2 VM but I didn't consider myself to be part of their community (I have since switched to Amazon Linux, for what it's worth).

So all that Red Hat did was basically restructure their collaboration with downstream distros. On one hand they enabled those distros to collaborate even more to RHEL development, which is now public (including individual patches to the kernel, if you remember the circa 2011 kerfuffle). On the other hand release rebuilds are entirely in the hand of the community.

Now, I am not saying everything was perfect. The announcement sucked in many ways, and there still isn't a good solution to use RHEL container images on public CI. People inside Red Hat (including me) will all tell you the same. However, it's intellectually dishonest to ignore that there was and is a CentOS world that goes beyond "I need free Linux and I don't/cannot use Debian", and Red Hat has been very receptive to the needs of that world.

Re: Rocky Linux 9.0

#106

Post by freedomben »

gh02t wrote: post #44
freedomben wrote: post #38

Earlier quoted context omitted.

IBM had nothing to do with the CentOS decision. It was long time Red Hat people who made the decision. I don't agree with everything about the decision, but I don't think it's as bad as most people say it is[1]. You are definitely right that RH has pushed things in their own interest, but if those things don't offer value to the broader community, then the community won't adopt them. Red Hat can't force Debian or Ubu…

While I agree with you, it is fair to say RH has an unusual amount of leverage with respect to forcing things. They directly control a lot of big ticket projects, and have powerful leadership positions in others. They can coordinate major changes across the board and the momentum they can throw behind some decisions can certainly exert a LOT of pressure. This isn't necessarily a bad thing though - as you note it can…

Agreed!

Re: Rocky Linux 9.0

#107

Post by freedomben »

js4ever wrote: post #43
tannhaeuser wrote: post #33

Earlier quoted context omitted.

I'd say it's a two-sided sword. There's no question RH employs/founds large parts of Linux development, with only Suse being remotely as involved (maybe historically). OTOH, RH has pushed "innovations" such as systemd purely in their own interest, fragmenting a once-strong and user-centric F/OSS Unix community also including the BSDs into a Linux-only cloud slavedom. Plus, it was IBM/RH who cancelled the CentOS roadm…

They also created quay.io and fragmented docker global public repo by asking sponsored FOSS project to publish ONLY on quay. I really dislike redhat and won't touch anything related with them. I really don't understand why companies would pay thousands per server for support. I'm working with Linux systems since 2 decades and never needed to pay a cent in licenses or support.

Red hat didn't create quay, they acquired it. But even if they had made it, I think that's an important and healthy thing to have. You could call it fragmentation, but I see it as taking on a monopoly by offering choice. Docker fully supports alternative registries, so I wouldn't consider that fragmentation either. Remind me a lot of when Apache was the only http server, and nginx came out. Some people said fragmentation, but most said, "cool, we have a choice and some market competition, without which development tends to stagnate"

There are things to dislike about red hat, but quay is one I would praise them for instead.

Re: Rocky Linux 9.0

#108

Post by bonzini »

josephcsible wrote: post #102
bonzini wrote: post #99

Earlier quoted context omitted.

BZ number for the third? Also what components?

The components for the first two were NSS (Network Security Services, not Name Service Switch) and GCC, and the third was Kerberos. I don't have the BZ number handy right now.

For the first two it's probably that they were considered not bad enough for an asynchronous release. With CentOS Stream, you would have a better understanding of when the patch was committed and more certainty of whether it would be fixed in the next minor (6-months cadence) release.

Feel free to write to me at pbonzini@redhat.com, I don't work on Kerberos myself but I know people in the team.

Re: Rocky Linux 9.0

#109

Post by Elv13 »

awill wrote: post #94
wolverine876 wrote: post #92

Earlier quoted context omitted.

> Note that AlmaLinux released 9.0 about 40 days ago. They're also significantly faster with releasing minor updates (including releases like 8.6). 'Faster' is not a recommendation for an OS developer, IMHO; I don't need operating systems faster - in fact, I like to give them time to mature in other people's hands and to test them myself. The only things I want ASAP are security fixes that are critical to my systems.…

Why is this a concern? If Alma and Rocky are essential identical, speed is the only/main differentiator, and it matters. If an urgent vulnerability gets patched in RHEL, and I'm using Rocky or Alma, I want the one that gets patched sooner. Right now, that's Alma.

Would you rather have `.rpm`s build by someone on the command line using `rpmbuild` or `mock` or something built by a CI with proper bootstraping and "nearer" to reproduceable builds. Also, by cutting corners on the CI, you risk introducing mild ABI problems which wont crash, but causes instabilities and potential security vulnerabilities. If they say they needed time to do it properly, I respect that.
Post reply on HN