"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
Agree. A public announcement of a decision on this, when made, would help head off a lot of user questions.
Phil
P.S.
Apologies to the QA list for the cross-post. On refection I thought a more public forum would be appropriate for this discussion.
Phil Schaffner writes:
"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
Agree. A public announcement of a decision on this, when made, would help head off a lot of user questions.
+1, 5.6 should have higher prio.
Phil
P.S.
Apologies to the QA list for the cross-post. On refection I thought a more public forum would be appropriate for this discussion.
+1, not everyone uses twitter
CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
On Thu, Jan 13, 2011 at 10:47 AM, nux@li.nux.ro wrote:
Phil Schaffner writes:
"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
Agree. A public announcement of a decision on this, when made, would help head off a lot of user questions.
+1, 5.6 should have higher prio.
+1 for 5.6. I think 5.6 is more important than 6.0 at this point. Many sysadmins might even wait for 6.1 (avoiding .0 releases). Security fixes in 5.6 should be given a higher priority. Also demand (request) for php5.3 has been so high ...
Akemi
Phil Schaffner wrote:
"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
Agree. A public announcement of a decision on this, when made, would help head off a lot of user questions.
Phil
Let me express my personal opinion on that discussion (there was already a small discussion about that on IRC). I'd prefer seeing 5.6 than 6.0 and for specific reasons :
- the centos 5.x install base is there while there is (obviously) no centos 6 install base. - So those people having machines in production, faced to the net, etc, etc would prefer having their machines patched and up2date (security first !) - people running CentOS 5.x on servers and waiting for php53 packages, now officially included - on the build side, the el5 build process is clearly identified and known since 2007 : packages with branding issues are already identified and patches/artwork is already there, meaning that it will be probably (surely) faster to have 5.6 out of the door than 6 - same rule for the QA process : people from the QA team can blindly focus on their previous tests, and just have a look eventually at some newer packages (a few, like php53 but not that much in comparison with el6)
On 01/13/2011 05:31 PM, Phil Schaffner wrote:
"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
Agree. A public announcement of a decision on this, when made, would help head off a lot of user questions.
From my perspective:
- 5.6 impacts existing installs
- 5.6 and updates contain security issues
- 5.x buildsys and process is fairly well tested and completely independent from 6.x; the main resources are not even hosted in the same country!
- the 5.6 testing process, for whatever automation we have at the moment, is completely independent from 6.x's stuff ; the patching process is also completely automated ( Except for a few packages that I need to do by hand and a handful others that pitch in with specific patches - but overall its fairly quickly done )
- Finally, all packages for 5.6 are in the buildqueue and first round of builds is complete ( the last package dropped out about 2 minutes back ).
The only place where there is an overlap is in the QA process. But my guess is that having been through 5 previous releases, the QA guys can address this fairly quickly.
however there are a few other implications that need to be considered before making a decision. Firstly, the mirror situation. If we were to drop 5.6 in a few days time - we would need to wait for mirror's to stabilise before we move to releasing 6.0. I am not sure what that time lag needs to be, but it should be possible to workout.
What I am going to propose is that lets let both the threads run through for the moment. Lets meet in #centos-devel@irc.freenode.net at 1600 hrs on Friday the 14th Jan 2011 and talk about this, get a plan together. That time should hopefully be late enough for the Americans to be awake, but early enough for the Germans to not be drunk yet :)
- KB
On Thu, Jan 13, 2011 at 10:30:55PM +0000, Karanbir Singh wrote:
From my perspective:
- 5.6 impacts existing installs
- 5.6 and updates contain security issues
This alone seems enough. The rest is frosting. :)
Matthew Miller wrote:
On Thu, Jan 13, 2011 at 10:30:55PM +0000, Karanbir Singh wrote:
From my perspective:
- 5.6 impacts existing installs
- 5.6 and updates contain security issues
This alone seems enough. The rest is frosting. :)
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are actually a couple of years late. How much has changed form 5.5 + updates to 5.6. How many existing systems will use the iso's instead of yum update. Iso's are primarily used for new installs. If I am making a new install, am I waiting for 5.6 or for 6.x? I had to leave CentOS for many of my systems a couple of years ago because it did not support the newer applications. So 6 fills a void currently painfully handled by Fedora instead of an enterprise class system. Bug fixes are needed by installed systems, they should be released as soon as the bug is fixed. Point changes are primarily a snapshot taken to speed up an install on a new system not to update a current system. Hubert
On Thu, Jan 13, 2011 at 10:59:10PM -0600, Hubert Bahr wrote:
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are
Except for the whole "supported until 2014" thing.
actually a couple of years late. How much has changed form 5.5 + updates to 5.6. How many existing systems will use the iso's instead of
Many will for new installs / rollouts where network installs, cobbler, etc are not available to them.
yum update. Iso's are primarily used for new installs. If I am making a new install, am I waiting for 5.6 or for 6.x? I had to leave CentOS for many of my systems a couple of years ago because it did not support the newer applications. So 6 fills a void currently painfully handled by Fedora instead of an enterprise class system. Bug fixes are needed
Third-party repos provide *MUCH* of the functionality that Fedora provides; and don't bring the rolling target issues with them.
by installed systems, they should be released as soon as the bug is fixed. Point changes are primarily a snapshot taken to speed up an install on a new system not to update a current system.
People are going to continue to install 5 until 5 is EOL. That is still 4+ years out. And as such they need ISO sets available.
People, at least 2.5 million if the ip count that Karanbir has mentioned is accurate, use C5; expecting that all of them are going to leave 5 behind and use 6 is just silly. *Many* people are locked into a platform for the life of that platform. These are *existing* users, not those planning on migrating to 6 whenever it is available.
John
John R. Dennison wrote:
On Thu, Jan 13, 2011 at 10:59:10PM -0600, Hubert Bahr wrote:
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are
Except for the whole "supported until 2014" thing.
actually a couple of years late. How much has changed form 5.5 + updates to 5.6. How many existing systems will use the iso's instead of
Many will for new installs / rollouts where network installs,
Point made new installs/rollouts
cobbler, etc are not available to them.
yum update. Iso's are primarily used for new installs. If I am making a new install, am I waiting for 5.6 or for 6.x? I had to leave CentOS for many of my systems a couple of years ago because it did not support the newer applications. So 6 fills a void currently painfully handled by Fedora instead of an enterprise class system. Bug fixes are needed
Third-party repos provide *MUCH* of the functionality that Fedora provides; and don't bring the rolling target issues
Painful is fedora, but hardware/new apps with 5 not filling the void. Upstream recognizes the loss to other vendors so released 6.0 well before 5.6 although 5.6 is much easier.
with them.
by installed systems, they should be released as soon as the bug is fixed. Point changes are primarily a snapshot taken to speed up an install on a new system not to update a current system.
People are going to continue to install 5 until 5 is EOL. That is still 4+ years out. And as such they need ISO sets available.
Agreed but updates full fill the major majority of the needs for the current installs. I never advocated dropping 5 just keeping the release order in the same sequence as upstream.
People, at least 2.5 million if the ip count that Karanbir has mentioned is accurate, use C5; expecting that all of them are going to leave 5 behind and use 6 is just silly.
This statement was asinine since nobody expects systems "satisfied" by C5 to switch. But do not expect those dissatisfied by C5 to wait while you switch the release order of the upstream vendor.
*Many* people are locked into a platform for the life of that platform. These are *existing* users, not those planning on migrating to 6 whenever it is available.
John
CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
On 1/13/11, Hubert Bahr hab@hbahr.org wrote:
John R. Dennison wrote:
On Thu, Jan 13, 2011 at 10:59:10PM -0600, Hubert Bahr wrote:
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are
Except for the whole "supported until 2014" thing
actually a couple of years late. How much has changed form 5.5 + updates to 5.6. How many existing systems will use the iso's instead of
Many will for new installs / rollouts where network installs,
Point made new installs/rollouts
Non-sequitur, your facts are uncoordinated. Check back again when you think about more than one use case.
jerry
On Thu, Jan 13, 2011 at 11:56:30PM -0600, Hubert Bahr wrote:
Painful is fedora, but hardware/new apps with 5 not filling the void. Upstream recognizes the loss to other vendors so released 6.0 well before 5.6 although 5.6 is much easier.
I will give you the hardware edge here as it's a valid argument; but the I still maintain that most "modern" userland functionality can be had by appropriate use of third-party vetted repos.
Agreed but updates full fill the major majority of the needs for the current installs. I never advocated dropping 5 just keeping the release order in the same sequence as upstream.
C6 is already quite late, for various definitions of "late"; moving it back however much more to get 5.6 dealt with and out the door and off the dev's plates isn't going to make or break it for most I wouldn't think. Speaking for myself, I want 5.6 and have little use for 6 at the present time. 5.6+security rollups is a much higher priority for myself and my clients.
This statement was asinine since nobody expects systems "satisfied" by C5 to switch. But do not expect those dissatisfied by C5 to wait while you switch the release order of the upstream vendor.
Those that aren't satisfied are free to purchase appropriate upstream entitlements :)
John
Hubert Bahr hab@hbahr.org писал(а) в своём письме Fri, 14 Jan 2011 07:59:10 +0300:
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are actually a couple of years late. How much has changed form 5.5 + updates to 5.6. How many existing systems will use the iso's instead of yum update. Iso's are primarily used for new installs. If I am making a new install, am I waiting for 5.6 or for 6.x? I had to leave CentOS for many of my systems a couple of years ago because it did not support the newer applications. So 6 fills a void currently painfully handled by Fedora instead of an enterprise class system. Bug fixes are needed by installed systems, they should be released as soon as the bug is fixed. Point changes are primarily a snapshot taken to speed up an install on a new system not to update a current system. Hubert
If you can afford Fedora on your systems, there isn't need to worry about C6, because you've already given up on stability. And those who want stability, still use C5.
Alex/AT wrote:
Hubert Bahr hab@hbahr.org писал(а) в своём письме Fri, 14 Jan 2011 07:59:10 +0300:
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are actually a couple of years late. How much has changed form 5.5 + updates to 5.6. How many existing systems will use the iso's instead of yum update. Iso's are primarily used for new installs. If I am making a new install, am I waiting for 5.6 or for 6.x? I had to leave CentOS for many of my systems a couple of years ago because it did not support the newer applications. So 6 fills a void currently painfully handled by Fedora instead of an enterprise class system. Bug fixes are needed by installed systems, they should be released as soon as the bug is fixed. Point changes are primarily a snapshot taken to speed up an install on a new system not to update a current system. Hubert
If you can afford Fedora on your systems, there isn't need to worry about C6, because you've already given up on stability. And those who want stability, still use C5.
If C5 doesn't cut the mustard you are forced to a different solution. Lack of stability in Fedora is a Pain thus the need for 6. Obsolescence is also a pain not all applications have the same demands. You are lucky if your demands are met by less than near state of the art. Of course hardware is evolving so fast and new designs are quite often provide much more bang for the buck that it is uneconomical to maintain them. We used to buy source code to maintain systems far past their prime, but maintenance costs soon drowned out replacement costs, so strategies evolved. So to must the OS. I would prefer not to use two different OS's but I am updating hardware. CentOS 6 is so late that I already have developed my own binaries and will use them. No I can not distribute them. Even with this solution it will take time to migrate the remaining systems. EPEL has a schedule to release EPEL 6 that completely eliminates the use of Fedora, I am very glad to be able to get back to an Enterprise system. Nobody has yet told me what the difference is between 5.5 + updates and 5.6 that makes it essential to push back the rollout of 6.0. I have no argument with security fixes, but aren't those already covered by the updates? Are the new features so essential to the installed base that they need them before the rollout of C6? How long a delay is that? No EPEL 5 etal does not meet my needs for all of my systems although most still use it. I have avoided jumping to non-rpm based distributions, but the temptation is still there. On some of my systems they are the easy way out due to specific applications which are not covered by epel etal supplements. Yes I have already purchased some subscriptions, I would have preferred to donate the money to CentOS. It is now spent so it is no longer available for donation. Don't slam a different perspective, please try to understand it. Hubert
CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
On Fri, Jan 14, 2011 at 01:24:43AM -0600, Hubert Bahr wrote:
If C5 doesn't cut the mustard you are forced to a different solution.
Yep, C5 base/updates and your 3rd-party repo of choice for userland and elrepo for drivers for more modern hardware in many cases.
Nobody has yet told me what the difference is between 5.5 + updates and 5.6 that makes it essential to push back the rollout of 6.0. I have
php53 and bind97 are show-stoppers for many; both are obtainable outside the official channels now, but having them as part of base is quite useful for many. bind97, with the support for DNSSEC it offers, is a hard requirement in many industry sectors at the present time.
no argument with security fixes, but aren't those already covered by the updates? Are the new features so essential to the installed base that they need them before the rollout of C6? How long a delay is that? No
Yes, in many peoples opinions they are quite essential.
EPEL 5 etal does not meet my needs for all of my systems although most still use it. I have avoided jumping to non-rpm based distributions, but the temptation is still there. On some of my systems they are the easy way out due to specific applications which are not covered by epel etal supplements.
epel is not the only game in town.
Yes I have already purchased some subscriptions, I would havepreferred to donate the money to CentOS. It is now spent so it is no longer available for donation.
The project hasn't taken monetary donations since '09 as far as I know. You are still supporting the project, albeit indirectly, by getting RH entitlements. Supporting the upstream is always in the project's best interest.
Don't slam a different perspective, please try to understand it.
Yep, please try to see the point that 5.6 is extremely important to many.
John
Am 14.01.2011 08:24, schrieb Hubert Bahr:
Nobody has yet told me what the difference is between 5.5 + updates and 5.6 that makes it essential to push back the rollout of 6.0.I have no argument with security fixes, but aren't those already covered by the updates?
No. 5.6 contains fixes that are not available for CentOS until 5.6 is out. So it's not about new installs, it's about keeping existing 5.x installations secure.
fs
Hi Hubert,
On 01/14/2011 04:59 AM, Hubert Bahr wrote:
From my perspective:
- 5.6 impacts existing installs
- 5.6 and updates contain security issues
A different perspective. RHEL-5, CentOS-5 no matter what point change is essentially obsolete. Thus the need for RHEL-6, CentOS-6 which are actually a couple of years late. How much has changed form 5.5 +
That makes perfect sense - however, its not a case of one or the other, its just a case of which one we get out of the door first. Also, since the QA team is largely a different set of people helping out with CentOS-6 and the infra team, there might be a a bit of potential to keep more people busy productively through the next few weeks if we were to get the 5.6 isos and tree over into QA first.
Lets see if we can get a decision on irc later today. The over whelming requests seem to be to get 5.6 out first, but its always good to talk about - there are valid points in either direction.
- KB
Dne 13.1.2011 23:30, Karanbir Singh napsal(a):
however there are a few other implications that need to be considered before making a decision. Firstly, the mirror situation. If we were to drop 5.6 in a few days time - we would need to wait for mirror's to stabilise before we move to releasing 6.0. I am not sure what that time lag needs to be, but it should be possible to workout.
What I am going to propose is that lets let both the threads run through for the moment. Lets meet in #centos-devel@irc.freenode.net at 1600 hrs on Friday the 14th Jan 2011 and talk about this, get a plan together. That time should hopefully be late enough for the Americans to be awake, but early enough for the Germans to not be drunk yet :)
- KB
Karanbir, so the C5 process is fairly clear and should be simple and quick. 21 days to release might be enough and I guess now it the time to internally set some deadline. 1st week - build process 2dn week - QA/rebuild iteration 3rd week - seeding the mirrors
One more thing, I can remember a lot of people willing to help don't want to use IRC.
Regrads, DH
On 01/14/2011 07:23 AM, David Hrbáč wrote:
so the C5 process is fairly clear and should be simple and quick. 21 days to release might be enough and I guess now it the time to internally set some deadline. 1st week - build process
For c5, that might be more than whats needed - but lets assume that things take a week to resolve because we are also working with the c6 package. We should be able to get an iso set for 5.6.build1
2dn week - QA/rebuild iteration
This is the bit that's hard to put a time on, the QA guys dont seem to like being time box'd in; having said that most of them seem to agree that since their own process are well rehearsed at this time, it should be a lot faster than the c6 stack.
One more thing, I can remember a lot of people willing to help don't want to use IRC.
That is an issue for the time being, using irc is just so much quicker and more productive at the moment. Ideally, getting some sort of an webapp together that lets people coordinate in an async manner would be the way to go, but having tried quite a few of them we didn't really come up with anything that we could use out of the box. I know that Steve has been working on the open-atrium install, but it might still be a few months before we can use it for a proper release yet.
- KB
On 01/13/2011 10:30 PM, Karanbir Singh wrote:
What I am going to propose is that lets let both the threads run through for the moment. Lets meet in #centos-devel@irc.freenode.net at 1600 hrs on Friday the 14th Jan 2011
that was meant to be 16:00 UTC
- KB
As per discussion on mailing list if some more hardware is helpful I have a DL380G6 empty of anything right now in my rack I can rebuild and/or spin VMs on for ISO/anaconda/anything testing.
It doesn't have a public address space but I can provide feedback at the very least.
James Hogarth
On 13/01/11 19:31, Phil Schaffner wrote:
"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
My vote goes for 6.0. We can live without 5.6, but 6.0 is really needed.
Whatever.
But in my opinion, both are important, but I think CentOS 5.6 is more important than 6 at this time. Yes, I'm really expecting for CentOS 6, but when C6 is ready I will take some time to migrate my "old" CentOS 5 platform to C6. Deploy a new C6 server will be really easy, but migrating my "outdated" systems to them, don't.
In other side I can easily update any CentOS 5.x to CentOS 5.6 (I still have some Cent OS 5.3 running mission critical things), and that's give me more time to do a migration plan.
I still think those people who waited RHEL6/C6 for so long time would wait a month more or two. Because the people who wants fresh packages shouldn't use Cent OS. I'm always looking for stability in first place, and the C5 base is very well consolidated in hardware suppliers and software companies.
(and thanks for the email in the list, I don't use twitter).
2011/1/14 Veiko Kukk veiko@ekp.ee
On 13/01/11 19:31, Phil Schaffner wrote:
"So: QA 5.6 and 6.0 in parallel or prefer one release over the other. Imho, 5.6 impacts existing installs, should get higher pref. Thoughts ?"
My vote goes for 6.0. We can live without 5.6, but 6.0 is really needed.
-- Veiko _______________________________________________ CentOS-devel mailing list CentOS-devel@centos.org http://lists.centos.org/mailman/listinfo/centos-devel
-
Akemi Yagi -
Alex/AT -
David Hrbáč -
Fabian Arrotin -
Felix Schwarz -
Hubert Bahr -
James Hogarth -
Jerry Amundson -
John R. Dennison -
Karanbir Singh -
Lucas Timm LH -
Matthew Miller -
nux＠li.nux.ro -
Phil Schaffner -
Veiko Kukk