As a random aside, I’m installing two new severs for Virtualmin.com next Monday or Tuesday, and it appears they’ll be running Scientific Linux (since there is no CentOS 6, and we require some of the latest stuff to really get the most out of Cloudmin and some of the new features we’re developing). So, barring a miraculous CentOS 6 release sometime between now and next Monday, you’ll see official Scientific Linux 6 support in the Virtualmin install script sometime next week. Cloudmin might already install there, I’m not sure…if it doesn’t, it’ll also get support next week.
It won’t be kept secret. When it works, it’ll be listed in the OS Support page, and I’ll post to the News forum about it. I’ve been poking at hardware issues (drivers and BIOS problems with our new servers), so I haven’t done anything with Virtualmin or Cloudmin on them yet. Since it didn’t get done last week, everything was delayed for several days by Texas Linux Fest.
I know you are working on SL 6 but does virtualmin currently work on SL 5 as an A-supported OS, and would a Centos 5 backup of a virtual server restore properly on SL 5 and also SL6 when available? I am currently using php 5.2.17 from bleed.
So, barring a miraculous CentOS 6 release sometime between now and next Monday,
you’ll see official Scientific Linux 6 support in the Virtualmin install script sometime next week
As of 2011-04-15, centos-devel said “A build and QA plan is being worked out now with the QA team”, .
Figure that means about 4 weeks would be my guess. On 2011/4/14, developers hinted that around May10th they
might send it over to the quality assurance team, who would do their thing, then it would be 48 to populate the mirrors.
So maybe around the end of May based on that.
I have also gone ahead and installed a new server using SL6, which I understand could be easily
converted to CentOS 6 later if desired. If any testing or such is desired on SL6 I would be
glad to help in any way I can. I do have 15 years of experience on RPM based servers, and
I am a programmer, so I might be able help. It has been several years since I mucked about
with the Virtualmin internals, though.
A bit off-topic: but why would one switch from SL to CentOS? SL has a far bigger base of devs/mataintainers. In addition it comes with a few extra features compared to CentOS. So if you don’t need the 0.5% additional binary-compability, there seems to be no good reason for this step.
I would appreciate the opinion of an experienced admin.
CentOS is used in a very large percentage of hosting deployments, while SL is pretty much non-existent in the hosting space (with a name like Scientific Linux, I can see how very few people would consider it for web hosting). CentOS has a much larger userbase than SL, in general, as well.
In short, CentOS is the standard in the hosting industry. Virtualmin is used for web hosting. Thus, CentOS is the most popular OS for use with Virtualmin.
As for switching…I don’t plan to switch our servers to CentOS, even after CentOS 6 arrives. As far as I can tell, Scientific Linux will work fine for what we need, so I would see no reason to switch. And, once we commit to supporting and OS, we support it for the entire supported lifecycle of the OS (so we support CentOS releases for 5 years). I’m not sure what the lifecycle of SL is; if it’s not as long as RHEL/CentOS, then I’d consider that a good reason for switching.
I appreciate your opinion which I basically share. I’d like to point out a few things based on your reply.
CentOS is well spread - absolutely. As far as I know has it been the only “real” and free RH clone - which might be the main (and only?) reason for the huge amount of CentOS installations.
SL on the other hand is “young”. CentOS started with version 3 while SL started with version 4 if my informations are correct.
As you’ve already pointed out, might the name be a little bit “unfortunate”. But how many distros are actually out there with a descriptive name? (and to be honest, when I read CentOS for the first time, my first thought was “cheap” :D). Scientific Linux has it’s name because it’s founded(?) and maintained primarely by scientific organisations like CERN and a bunch of other (educational) institutions.
In fact, this could lead us to another point: how much do we trust CentOS or SL?
Most people (me included) would tend to the community OS (as far as I know is SL community based too, but the sponsors have the last word). Anyway, the question is how we do define “trust”? Would you agree if I say, we’re talking about maintance and lifecycles?
Basically, Red Hat makes the rules. That leaves two questions:
how fast are patches/updates adapted?
are there (negative) differences regarding the lifecycles?
I don’t have any numbers, facts or proofs. But based on the fact that SL is used in (I suppose) critical enviroments, is my assumption that SL could be faster.
in my opinion, this is the primary point and source of the whole discussion.
a.) Why are we even talking about SL? Because CentOS seems to be… not actually dead, but as far as I know it’s simply unpredictable at the moment. As far as I’m informed, there is still no roadmap regarding version 6. (Don’t get me wrong. This is no criticism and I appreciate their work and their priceless contribution to the community). Fact is, they have a lack of developers and contributors (actually, what would you expect when you release a free version of a commercial product :D).
SL on the other hand has released version 6 this March and I would say they have enough manpower (and money?) to maintain their distro. Their website quotes the same lifecycles as the original does.
If we put all those pieces together I come to the following conclusion:
In my view SL seems to become the “new” CentOS
It might have a smaller user-base (yet) but the foundation looks more robust to me.
If you choose Linux as server OS it makes no difference which distro you are going to use (in my opinion are there only two distros I’d consider for a server: Debian and RedHat/CentOS/SL)
One has to decide if he trusts a distro/company/organisation or not. I’m personally very skeptical. In fact, the only company I trust is Red Hat (and only because of Greg Kroah-Hartmann). Since the main sponsors are CERN, Fermilab and ETHZ I would tend to trust SL (hey, CERN could blow up the world, why would they mess around with Linux? :D).
For any other concerns -> Red Hat
However: How about Red Hat 6 support for Virtualmin?
: but why would one switch from SL to CentOS? SL has a far bigger base of devs/mataintainers. In addition it comes with a few extra features compared to CentOS. So if you don’t need the 0.5% additional binary-compability, there seems to be no good reason for this step.
The two are virtually identical I understand, so it could come down to what you LABEL the system.
Why would you call it CentOS instead of Scientific Linux? Installing VirtualMin using the recommended
method, the automatic method, would be one reason. The auto installer will recognize CentOS as
a supported OS, but won’t recognize SL at this time. Similarly for other software or people who
may recognize CentOS but not SL. Proprietary drivers, for example, or maybe support purchased
for a database system or what not, where their policy people may not know what CentOS and SL are
pretty much identical.
I haven’t actually done significant work on my SL 6.0 system yet, so I’m not aware of whatever other
reasons may exist. That’s another reason I like knowing that I can easily switch it, actually, the fact
that I don’t know what I might run in to. I know that IF I run into some unexpected hurdle with SL updates
or whatever I can always “punt” by converting renaming it as a CentOS system, then running the updates.
The two are virtually identical I understand, so it could come down to what you LABEL the system. Why would you call it CentOS instead of Scientific Linux?
Well, although both Distros are trying to be as binary-compatible as possible with the original one (which in fact isn’t possible, because they will always lack the non-free Red Hat tools), they’re actually modifying/adding and - most importantly - removing packages.
That might (or might not) be a another reason to prefer one over the other. (i.e. i like the additional shipped repos in SL).
Anyway, the methos how VM identifies the used OS is a combination of the install-script, the RPM and a file on their webserver. Somewhere on my TODO-list is an entry for building a RPM which doesn’t care about the underlying OS and some of the dependencies (the admin ought knows what he’s doing).
However I’m very occupied lately. Additionally am I still optimistic, that there will be an official VM release with RH/CO/SL 6 support.
Btw. to be fair… CentOS released 5.6 while SL still stucks on 5.5 (purpose?)
Yeah CentOS decided to put 6.0 on hold to get 5.6 out the door, while SL decided to do 6.0 first, then 5.6.
For official RH/CO/SL 6 support, expect CentoS 6.0 in about two weeks - roughly June 1st or 2nd.
Because CentOS seems to be..... not actually dead, but as far as I know it's simply unpredictable at the moment. As far as I'm informed, there is still no roadmap regarding version 6. (Don't get me wrong. This is no criticism and I appreciate their work and their priceless contribution to the community). Fact is, they have a lack of developers and contributors (actually, what would you expect when you release a free version of a commercial product :D).
There IS a roadmap and a new open development platform at http://qaweb.dev.centos.org/ .
Recent developments have shown the need for changes to the methods and systems used for
CentOS in order to allow more volunteers to help more easily. The OpenAtrium platform is one
of several new systems being tested to open it up more so that more people can contribute.
Compared to other open source projects, SOME of the leaders of CentOS aren’t AS INTERESTED
in whether or not other people use or contribute to the project. Other leaders are interested in
opening it up so that more people can contribute and actions are being taken in that regard.
Something else which may be somewhat new is that the leadership is realizing that users had been
starving for information. They have recently begun t get the information out a little better, announcing
a while ago that they planned to send 6.0 to the QA team on a certain date, etc.
As you’ve mentioned, it seems like the information-flow is a bit confusing, but the QA blog looks like a good source for news. I’m surprised about the progress they made, the first Live CDs have been released for testing.
I’m looking forward for their released and the differences to SL.
About the “interested” thing… well, I know by myself how hard it is to maintain OSS projects (needlessly to mention the complexity of a whole distribution), especially when you’re a student or occupied with other private stuff. So I wouldn’t blame them - everybody has it own priorities. But that is one of the key-points imho in the discussion SL vs. cOS.
Anyway, let’s wait for the final and furthermore… Virtualmin support.