Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Gentoo Apache2 Config Change Idiocy
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, ... 9, 10, 11  Next  
This topic is locked: you cannot edit posts or make replies.    Gentoo Forums Forum Index Gentoo Chat
View previous topic :: View next topic  
Author Message
Monkeh
Veteran
Veteran


Joined: 06 Aug 2005
Posts: 1656
Location: England

PostPosted: Sun Sep 25, 2005 2:07 pm    Post subject: Reply with quote

macin wrote:
You could have at least linked to the upgrade instructions at the end of the ebuild.


Code:
if has_version '<net-www/apache-2.0.54-r30' && has_version '>=net-www/apache-2.0.0' ; then
                einfo "Configuration locations have changed, you will need to migrate"
                einfo "your configuration from /etc/apache2/conf/apache2.conf and"
                einfo "/etc/apache2/conf/commonapache2.conf to /etc/apache2/httpd.conf."
                einfo
                einfo "Apache now checks for the old configuration and refuses to start"
                einfo "if it exists. You must remove the old configuration first"
                einfo
                einfo "You should also at this time rebuild all your modules"
                einfo
                einfo "For more information, see"
                einfo "    http://www.gentoo.org/proj/en/apache-upgrading.xml"
                einfo
        fi
Back to top
View user's profile Send private message
Kloeri
Retired Dev
Retired Dev


Joined: 02 Sep 2002
Posts: 144

PostPosted: Sun Sep 25, 2005 2:26 pm    Post subject: Reply with quote

There's a lot of good reasons for the changes. I'll copy some of them verbatim below for those people too lazy to read the carefully written documentation at http://www.gentoo.org/doc/en/apache-upgrading.xml :)

* The configuration that came with Gentoo was dramatically different from the upstream configuration that most users expect
* Many modules used similar code, but all did things their own way
* Modules didn't have a configuration standard
* Some modules could support both versions of Apache, but the ebuilds didn't handle that
* Choices available in Apache were not available for Gentoo users (for example MPMs)

I'm sure most of you would agree that the above points are all important and needed fixing. The changes have been a work in progress the last year or so and the new apache layout was unmasked (moved into ~arch) more than 6 months ago after announcements to gentoo-dev and gentoo-web-user as well as Gentoo Weekly News (see http://www.gentoo.org/news/en/gwn/20050228-newsletter.xml)

When we finally stabled the new layout on September 18th we again took great care to announce this move - again hitting every related mailinglist + Gentoo Weekly News. Further more the ebuilds asks people to read the Upgrading document listed below.

We've also had fairly extensive documentation available for several months now covering most almost all needed information and common problems people might run into. See:
http://www.gentoo.org/doc/en/apache-upgrading.xml
http://www.gentoo.org/doc/en/apache-troubleshooting.xml
http://www.gentoo.org/doc/en/apache-developer.xml

We're also staffing an irc-channel (#gentoo-apache on irc://irc.freenode.net) and are quite happy to answer questions and help with Gentoo apache in general there as well as answering as many threads on forums.gentoo.org as possible.

I'd also like to mention that the old layout isn't going away anytime soon - we're committed to maintain that with security and other important updates for at least a few months. We haven't set a cutoff date yet but we're not trying to force people into updating overnight either.

For apache-1.x new style is >=1.3.33-r10 and for apache-2.x new style is >=apache-20.54-r30. That means masking these versions is an easy way to put off the update until you can better make the changes.

Finally, it's my personal opinion that anybody being surprised by these changes and running production servers is doing a poor job. You can't just expect to do all updates blindly and never hit any problems. Often times the problems will be minor but it's easy to have ssl broken by seemingly small updates or similar. Of course you're all testing updates on non-production servers before implementing them on production servers, right? :)
Back to top
View user's profile Send private message
macin
n00b
n00b


Joined: 28 Apr 2004
Posts: 50

PostPosted: Sun Sep 25, 2005 2:49 pm    Post subject: Reply with quote

Monkeh wrote:
macin wrote:
You could have at least linked to the upgrade instructions at the end of the ebuild.


Code:
if has_version '<net-www/apache-2.0.54-r30' && has_version '>=net-www/apache-2.0.0' ; then
                einfo "Configuration locations have changed, you will need to migrate"
                einfo "your configuration from /etc/apache2/conf/apache2.conf and"
                einfo "/etc/apache2/conf/commonapache2.conf to /etc/apache2/httpd.conf."
                einfo
                einfo "Apache now checks for the old configuration and refuses to start"
                einfo "if it exists. You must remove the old configuration first"
                einfo
                einfo "You should also at this time rebuild all your modules"
                einfo
                einfo "For more information, see"
                einfo "    http://www.gentoo.org/proj/en/apache-upgrading.xml"
                einfo
        fi


my fault - i reemerged apache2 since it didn't work anymore. i was in hope for some maybe accidently deleted config updates. but then there was (of course) no einfo msg anymore. and the first time it slipped by...

Kloeri wrote:
There's a lot of good reasons for the changes. I'll copy some of them verbatim below for those people too lazy to read the carefully written documentation at http://www.gentoo.org/doc/en/apache-upgrading.xml

agreed

Kloeri wrote:
Finally, it's my personal opinion that anybody being surprised by these changes and running production servers is doing a poor job. You can't just expect to do all updates blindly and never hit any problems. Often times the problems will be minor but it's easy to have ssl broken by seemingly small updates or similar. Of course you're all testing updates on non-production servers before implementing them on production servers, right?

well that IS of course true. but my point is that gentoo is used not only by ppl with production servers, ppl who are not subscribed to gentoo mailing lists...
maybe thats a missing portage feature. some kind of dangerous flag for ebuilds - indicating that there is probably manual work to be done. maybe with some apriori warning message explaining how come or linking to further reference.
Back to top
View user's profile Send private message
lokey
n00b
n00b


Joined: 30 Aug 2004
Posts: 33

PostPosted: Sun Sep 25, 2005 7:45 pm    Post subject: Thanks Gentoo Reply with quote

Thanks Gentoo!

I've been meaning to rewrite my entire custom apache setup again from scratch. All I needed was the motivation of an emerge which decides to create a wholly new config file (httpd.conf) and fill it full of B.S. values thus breaking EVERYTHING!!

I mean, i thought nothing would happen until I ran etc-update! Isn't that what it was designed for!??!

Thats great, I mean you should have seen the look on my face today when about a week after emerging the new apache I restart it and nothing works! Priceless..

You're going to have to try harder next time Gentoo, I still have a few paying customers left and this borkage only took about 20 minutes to fix (unlike the Qmail config fiasco a few months ago - I gotta hand it to you on this one, breaking the email system completely and silently - that takes talent!).

Thanks for honing my admin skillz to a razor sharp edge!

-One of the Fewer and Fewer sys-admins running Gentoo on a production system.



P.S. Thanks also for not mentioning the Gentoo Newsletter in AT ALL in the HANDBOOK or INSTALLATION GUIDE. I suppose now that more and more people will be reading the newsletter you will be releasing information on future borking in an even more obscure location. Perhaps scratched into a cave wall somewhere in Germany?

:evil:
Back to top
View user's profile Send private message
amne
Bodhisattva
Bodhisattva


Joined: 17 Nov 2002
Posts: 6377
Location: Graz / EU

PostPosted: Sun Sep 25, 2005 8:14 pm    Post subject: Re: Thanks Gentoo Reply with quote

lokey wrote:

P.S. Thanks also for not mentioning the Gentoo Newsletter in AT ALL in the HANDBOOK or INSTALLATION GUIDE.


It's mentioned directly on WWW.GENTOO.ORG?!
_________________
Dinosaur week! (Ok, this thread is so last week)
Back to top
View user's profile Send private message
Kloeri
Retired Dev
Retired Dev


Joined: 02 Sep 2002
Posts: 144

PostPosted: Sun Sep 25, 2005 8:46 pm    Post subject: Re: Thanks Gentoo Reply with quote

Quote:
I mean, i thought nothing would happen until I ran etc-update! Isn't that what it was designed for!??!

Thats great, I mean you should have seen the look on my face today when about a week after emerging the new apache I restart it and nothing works! Priceless..

You're going to have to try harder next time Gentoo, I still have a few paying customers left and this borkage only took about 20 minutes to fix (unlike the Qmail config fiasco a few months ago - I gotta hand it to you on this one, breaking the email system completely and silently - that takes talent!).

Thanks for honing my admin skillz to a razor sharp edge!

-One of the Fewer and Fewer sys-admins running Gentoo on a production system.


Ohh, I'm sure your customers are happy to pay for your amazing sys-admin skills.. Why are you testing upgrades on your customers production servers and not even bothering to read *any* documentation at all, much less the warnings that the ebuild outputs while upgrading?
Back to top
View user's profile Send private message
j-m
Retired Dev
Retired Dev


Joined: 31 Oct 2004
Posts: 975

PostPosted: Sun Sep 25, 2005 8:56 pm    Post subject: Re: Thanks Gentoo Reply with quote

lokey wrote:

-One of the Fewer and Fewer sys-admins running Gentoo on a production system.


The fewer "sysadmins" of your flavor, the better for customers and for Gentoo. I'm running Gentoo on a couple of webhosting servers, but unlike you, I bother with reading the countless announcements and docs and ask if I'm not sure. I even resorted to such stupidity like testing on another machine and also have a chroot setup for testing. Oh, what a waste of time. Why not test on production servers, it must be the right way(tm), apparently. :lol:

I hope your customers are happy. :P
Back to top
View user's profile Send private message
Monkeh
Veteran
Veteran


Joined: 06 Aug 2005
Posts: 1656
Location: England

PostPosted: Sun Sep 25, 2005 8:59 pm    Post subject: Reply with quote

I'm going to say it again, if this update causes you problems, it's your own damn fault. It's pretty easy to understand this: Do not upgrade anything on a production system without testing it elsewhere first and knowing EXACTLY what it does.
Back to top
View user's profile Send private message
j79zlr
Apprentice
Apprentice


Joined: 05 Dec 2004
Posts: 235
Location: Chicago, IL

PostPosted: Sun Sep 25, 2005 9:13 pm    Post subject: Reply with quote

And I'll say this again, if you are running Gentoo as a production server, you shouldn't, portage is NOT stable.
Back to top
View user's profile Send private message
j-m
Retired Dev
Retired Dev


Joined: 31 Oct 2004
Posts: 975

PostPosted: Sun Sep 25, 2005 9:22 pm    Post subject: Reply with quote

j79zlr wrote:
And I'll say this again, if you are running Gentoo as a production server, you shouldn't, portage is NOT stable.


Hmm... and now for something completely different :roll:
Back to top
View user's profile Send private message
gkmac
Guru
Guru


Joined: 19 Jan 2003
Posts: 323
Location: West Sussex, UK

PostPosted: Sun Sep 25, 2005 9:27 pm    Post subject: Re: Thanks Gentoo Reply with quote

lokey wrote:
I've been meaning to rewrite my entire custom apache setup again from scratch.
I used this opportunity to jump from Apache 1 to Apache 2 since I knew I had to reconfigure from scratch. Actually, "reconfigure from scratch" is a slight lie, since I did backup my /etc/apache directory well in advance.
Quote:
All I needed was the motivation of an emerge which decides to create a wholly new config file (httpd.conf) and fill it full of B.S. values thus breaking EVERYTHING!!
Strange, I found the defaults in httpd.conf to be very reasonable indeed. It did seem to have that familiarity from when I ran Apache on RedHat, meaning that those coming from other distros won't find it so alien and not have to trudge through a rather strange configuration that was cut in half as apache.conf and commonapache.conf

I use virtual hosts, and I really like the way that you can now do it as one file for each virtual host within the "vhosts.d" directory, as opposed to one long Vhosts.conf in the old days. I also like the "modules.d" one-for-each-module file approach as well.

Am I the only one not complaining?
Back to top
View user's profile Send private message
Monkeh
Veteran
Veteran


Joined: 06 Aug 2005
Posts: 1656
Location: England

PostPosted: Sun Sep 25, 2005 9:28 pm    Post subject: Re: Thanks Gentoo Reply with quote

gkmac wrote:
Am I the only one not complaining?


Nope. I'm not either.
Back to top
View user's profile Send private message
Simba
n00b
n00b


Joined: 08 Nov 2002
Posts: 60

PostPosted: Sun Sep 25, 2005 9:52 pm    Post subject: Reply with quote

loki99 wrote:
amne wrote:
Simba wrote:
an sms would be great..


Or a girl wearing almost nothing jumping out of a cake whispering it into your ear...


:lol:


but seriously, it would be great if the emerge give warning about the change AND ask the user if they want to
continue the upgrade because it will break anything. I was just lucky that I emerged apache on a test server.
but still it takes me about 10 min. till I realized that there is a big change in the configuration. I knew and I read about the
change some times ago, but I though >= apache-2.0.54-r30 is still masked, because I see it in
package.gentoo.org, it is still hard masked now. or do I misunderstand with hard mask in package.gentoo.org?

If emerge give a warning about the change and ask the user if they want to continue, then it is absolutly users
fault if they just ignore the warning, and they can't blame developer about it. but now, you can't just blame the
user that they don't read mailing list or any GWN. People expect emerge will not break something, at least
don't break an important server application. But after this case people will have lesser trust to emerge.
And that is a very bad PR for Gentoo.
Back to top
View user's profile Send private message
Kloeri
Retired Dev
Retired Dev


Joined: 02 Sep 2002
Posts: 144

PostPosted: Sun Sep 25, 2005 10:11 pm    Post subject: Reply with quote

Simba wrote:
I knew and I read about the change some times ago, but I though >= apache-2.0.54-r30 is still masked, because I see it in package.gentoo.org, it is still hard masked now. or do I misunderstand with hard mask in package.gentoo.org?


You're misunderstanding packages.gentoo.org. The new apache layout was unmasked in february and have been in ~arch use for 6 months now.
Back to top
View user's profile Send private message
lokey
n00b
n00b


Joined: 30 Aug 2004
Posts: 33

PostPosted: Mon Sep 26, 2005 1:20 am    Post subject: Reply with quote

Quote:

It's mentioned directly on WWW.GENTOO.ORG?!


Yes, I fully admit that it is my fault for not checking the gentoo.org main page every single day. Silly me, I thought it would be enough to read your installation documentation.


Quote:

Ohh, I'm sure your customers are happy to pay for your amazing sys-admin skills.. Why are you testing upgrades on your customers production servers and not even bothering to read *any* documentation at all, much less the warnings that the ebuild outputs while upgrading?


My fault again, for not realizing that something as simple as a package upgrade will completely change the way apache reads its config files without me doing an etc-update or being notified of the changes beforehand.


Whats next? change from linux to freebsd passwd files so nobody can login? Since /etc/master.passwd doesn't exist you can make that change without etc-update getting in the way! That would be great for a laugh! Can I make this a feature request?


Quote:

And I'll say this again, if you are running Gentoo as a production server, you shouldn't, portage is NOT stable.


Ok.. I'll be honest, it wasn't on a production server. Its a pseudo-development server where the customers realize that its gentoo and are cool with the system not being rock solid. You would have to be out of your mind to run gentoo on a production server.

I switched from Freebsd and slackware because Freebsd wasn't cutting edge enough and I got tired of recompiling everything by hand in Slackware.

Is there some kind of middle ground where some developer looking to make a name for himself doesn't try to re-arrange everything and tell me that I can and can't do with MY apache config files? Said apache configs being re-designed to compensate for a frankly stupid and clumsy design in the first place?

I think that the new way is better, the problem I have with it is two fold:

1. I had already designed my apache configs to work the way I wanted. Yours were installed on top of these with complete disregard for whatever was in place, thus breaking not only apache and existing virtual host definitions, but my backend scripts as well.

2. How is Gentoo ever going to be stable when something so simple as an LAMP setup can't survive an emerge?
Back to top
View user's profile Send private message
j79zlr
Apprentice
Apprentice


Joined: 05 Dec 2004
Posts: 235
Location: Chicago, IL

PostPosted: Mon Sep 26, 2005 1:49 am    Post subject: Reply with quote

lokey wrote:
I switched from Freebsd and slackware because Freebsd wasn't cutting edge enough and I got tired of recompiling everything by hand in Slackware.


I never get this comment, FreeBSD's ports are updated atleast as frequently if not more frequently, and ports are testing and released well ahead of Gentoo, when you consider everything stays soft masked for weeks. Take a look at the previous gnome release for example. How long untill it was stable in Gentoo? 2 months after release? It was in ports within a few days of release. The only thing FreeBSD really lacks is some desktop support, and that is really the third parties fault for not providing them, for example a working flash plugin, or adobe reader plugin, so they have to use linux compatibility, which noone will argue is as good as it running natively.

I agree with the slackware comment as well, I found myself compiling half of the stuff on my system anyways, when they put Gnome support into the hands of Dropline it was the last straw, I switched to Gentoo.

I would also like to say for the record that I really do like Gentoo, it is my OS on all of my workstations, but I would not and do not use it as a server.
Back to top
View user's profile Send private message
cboyd
n00b
n00b


Joined: 25 Jul 2005
Posts: 21
Location: Kansas City

PostPosted: Mon Sep 26, 2005 1:21 pm    Post subject: Reply with quote

Quote:
maybe thats a missing portage feature. some kind of dangerous flag for ebuilds - indicating that there is probably manual work to be done. maybe with some apriori warning message explaining how come or linking to further reference.


Well, in my opinion if you're running a portage emerge sync and then an update world as some sort of cron job, that's your option and not portage. Perhaps you should just cron emerge sync, do an emerge -pv and then output it to an email to you, then you can read it before you update world.
Back to top
View user's profile Send private message
llongi
Retired Dev
Retired Dev


Joined: 15 Apr 2004
Posts: 459
Location: Switzerland

PostPosted: Mon Sep 26, 2005 2:27 pm    Post subject: Reply with quote

lokey wrote:

Yes, I fully admit that it is my fault for not checking the gentoo.org main page every single day. Silly me, I thought it would be enough to read your installation documentation.


Good, since it is. There is no point in arguing it, and no one requires you to read it each day, the announcement was made on the 12th, and the stabilization occourred on the late 18th, so practically a week. If in a week you don't manage once to check the website of your used distribution, if yon don't read forums, if you don't subscribe to any mailinglist and if you don't watch your updates, and just expect to forever update blindly and all has to work, then you didn't really understand what administrating a system with responsibility means.

lokey wrote:

My fault again, for not realizing that something as simple as a package upgrade will completely change the way apache reads its config files without me doing an etc-update or being notified of the changes beforehand.


Yeah, because you actually were notified of changes, even the ebuild said "the config has changed, please be careful and update".
What should we do? Let your server reboot and change all index.html and index.php sites with a "OMG APACHE CHANGED!" message so you noticed? We really went out to all channels we could reach users with, if you do not care about them and don't read them, and even don't check what the ebuild says to you when emerging, well... your fault.

Quote:

And I'll say this again, if you are running Gentoo as a production server, you shouldn't, portage is NOT stable.

lokey wrote:

Ok.. I'll be honest, it wasn't on a production server. Its a pseudo-development server where the customers realize that its gentoo and are cool with the system not being rock solid. You would have to be out of your mind to run gentoo on a production server.


Again, not true... I have a server I use for myself, not production with customers (not yet), but I have anything on it: my personal sites with Apache, PHP, my MySQL/PostgreSQL databases and a mail-system that does all: POP3/IMAP/SMTP, filtering, etc..
In two years Gentoo I've never had a breakage that at the end I couldn't say "ok, I was warned it could happen, my fault" or simply "ok, it's my fault"... I've ran the new Apache on this server since it existed (January 2005, 9 months it was available to anyone!) and it always worked, the migration was not so difficult, the documentation is there and explains stuff very well. So I believe, as do many people, that Gentoo can also perfectly be used on a production server. I'm no extremist Gentooer that says "always only Gentoo!!! W000T!", I am perfectly fine with people using Debian or FreeBSD, they are great server systems and I run also a server on Debian, but Gentoo, if used with knowledge and care (two things a sysadmin should have), it can be a really great server system and does a good deal of the work for you (with Gentoo I mainly see USE-flags etc. as "the" bonus, CFLAGS and running anythin with -O3 -fomg-faster-faster is just the part of users you see more because they yell more. :) I always use -O2 and that's it, performance is not the main point of Gentoo, especially not on a server imo, but customization through Portage is, as well as tight security (see Hardened-Gentoo), up-to-date software and great support on forums/irc (remember we're all here volounteers)).

lokey wrote:

Is there some kind of middle ground where some developer looking to make a name for himself doesn't try to re-arrange everything and tell me that I can and can't do with MY apache config files? Said apache configs being re-designed to compensate for a frankly stupid and clumsy design in the first place?


The design was wrong in the first place, or not really well tought, that we admit, and we fixed it. So you actually should be happy we righted a wrong situation, tested it for months and released it with announcements, documentation et all. If you ignore that, then there is not much to do, the change needed to be done, you say so yourself by admitting the old config was not good, so it was done and succesfully so!

lokey wrote:

I think that the new way is better, the problem I have with it is two fold:

1. I had already designed my apache configs to work the way I wanted. Yours were installed on top of these with complete disregard for whatever was in place, thus breaking not only apache and existing virtual host definitions, but my backend scripts as well.


Ok, the next time we will ask each single user of Apache on Gentoo how exactly he setup his own server, how he administers it and make a custom script for him that will take care of moving the configuration and all correctly, because we have hundredthousand devs just waiting for the chance to fix stuff you made on your system on your free choice... No seriously, you do your own stuff, no one knows about this, how do you expect now that we made it so that for everyone upgrading it would break nothing?
Btw, well designed backend scripts are fixed for the new situation in like 30 minutes, just change the path definitions and you're done. ;)

lokey wrote:

2. How is Gentoo ever going to be stable when something so simple as an LAMP setup can't survive an emerge?


Gentoo is stable, it really is, but only if used with the knowledge of what you're doing, why and what it will cause, what will happen etc.!
There are hundreds of users out there who managed to upgrade without problems, most are actually glad of the changes, and LAMP was always stable in the sense of "it works", if the configuration changes and you're informed about it, as you were, again: it's your fault for not caring/noticing and then complaining when it broke...

Best regards, CHTEKK.
_________________
Best regards, Luca.
Back to top
View user's profile Send private message
96140
Retired Dev
Retired Dev


Joined: 23 Jan 2005
Posts: 1324

PostPosted: Mon Sep 26, 2005 6:12 pm    Post subject: Reply with quote

lokey wrote:
Ok.. I'll be honest, it wasn't on a production server. Its a pseudo-development server where the customers realize that its gentoo and are cool with the system not being rock solid. You would have to be out of your mind to run gentoo on a production server.

*gasp* You mean, like all the official Gentoo servers that maintain all aspects of the distribution? Heaven forbid!

So, if the Gentoo devs are out of their minds for doing their development on Gentoo boxes, then you, as a user of Gentoo (and their software contributions), have inherited the "out of your mind" eclass. :P
Back to top
View user's profile Send private message
Akhouk
Guru
Guru


Joined: 23 May 2003
Posts: 476
Location: The Two Niles, Africa

PostPosted: Mon Sep 26, 2005 6:40 pm    Post subject: Reply with quote

I wasn't bitten by this one because I knew it was going to happen because I read the GWN. I planned the upgrade, tested it on a development server before running the upgrade out on my live servers. Everytthing went smoothly because I was prepared.

However, it is clear from this post that many people were stung because they don't read GWN, forums, or even the ebuild info when they upgrade. It is, IMHO, far to easy to upgrade a package (by doing emerge -up world) and totally miss import einfo messages.

I propose that packages that need dramatic changes in their upgrade have some ebuild restriction that prevents them being upgraded without a special command line option e.g. --really-upgrade. Without the flag it would stop and print the ewarn messages about what the upgrade is going to do.

Of course, such a flag shouldn't be routinely used but only used in special case upgrades like for this apache. Thus forcing people to read what is going to happen before it is too late.
_________________
AMD 64 3500+, 2Gb RAM DDR400, 2 x 180Gb SATA, 256Mb Nvidia
Back to top
View user's profile Send private message
grassushi
n00b
n00b


Joined: 11 Feb 2005
Posts: 54

PostPosted: Mon Sep 26, 2005 6:52 pm    Post subject: Reply with quote

I'll second that proposal.
Back to top
View user's profile Send private message
pablo_supertux
Advocate
Advocate


Joined: 25 Jan 2004
Posts: 2594
Location: Somewhere between reality and Middle-Earth and in Freiburg (Germany)

PostPosted: Mon Sep 26, 2005 7:13 pm    Post subject: Reply with quote

Akhouk wrote:
I wasn't bitten by this one because I knew it was going to happen because I read the GWN. I planned the upgrade, tested it on a development server before running the upgrade out on my live servers. Everytthing went smoothly because I was prepared.

However, it is clear from this post that many people were stung because they don't read GWN, forums, or even the ebuild info when they upgrade. It is, IMHO, far to easy to upgrade a package (by doing emerge -up world) and totally miss import einfo messages.

I propose that packages that need dramatic changes in their upgrade have some ebuild restriction that prevents them being upgraded without a special command line option e.g. --really-upgrade. Without the flag it would stop and print the ewarn messages about what the upgrade is going to do.

Of course, such a flag shouldn't be routinely used but only used in special case upgrades like for this apache. Thus forcing people to read what is going to happen before it is too late.


even so, most of sysadmin that don't read information would use --really-upgrade all the time and the same thing would happen.

I think there is a better solution (I'm afraid this is more a joke) if such ebuild would ask the user something. I mean, the ebuild can only resume, if the sysadmin answers a question correctly. The question should be taken randomly so that the sysadmin read the whole information. :lol:
_________________
A! Elbereth Gilthoniel!
silivren penna míriel
o menel aglar elenath,
Gilthoniel, A! Elbereth!
Back to top
View user's profile Send private message
96140
Retired Dev
Retired Dev


Joined: 23 Jan 2005
Posts: 1324

PostPosted: Mon Sep 26, 2005 7:15 pm    Post subject: Reply with quote

Akhouk wrote:
Of course, such a flag shouldn't be routinely used but only used in special case upgrades like for this apache. Thus forcing people to read what is going to happen before it is too late.

I dunno if that's really a feasible (or even good) idea.

Yes, Apache is now done a little differently. But really, it's nothing compared to the baselayout changes of the last several months--I think the first change was arguably the worst; the one that started splitting all the various configs into many, many smaller files in different locations. Granted, now /etc/conf.d/ is being used as it should, but I remember dispatch-conf came up with something like a hundred new config files to review.

If a new emerge flag was to be implemented, it would be most useful for baselayout changes (which have occured far too frequently since about March or April), and for packages that require more than, say, 20-50 file changes/additions.

How many config changes does the new Apache need, anyway?
Back to top
View user's profile Send private message
amne
Bodhisattva
Bodhisattva


Joined: 17 Nov 2002
Posts: 6377
Location: Graz / EU

PostPosted: Mon Sep 26, 2005 7:37 pm    Post subject: Reply with quote

emerge -pvul already shows you the ChangeLog of the new version you're going to install. Sure, there's a lot of boring stuff in there, but it should give you a general clue of what to expect.
Personally i'd like portage to interfere as little as possible with me by default, which means i don't want it to ask me anything (e.g. if i read and understood the warning). If i'm willing to see what happens, i'll do a emerge -uavl world, if i already did that on my other boxes and just want to upgrade my mp3 playing machine a blind emerge -u world is enough.
_________________
Dinosaur week! (Ok, this thread is so last week)
Back to top
View user's profile Send private message
sheldonh
n00b
n00b


Joined: 14 Sep 2004
Posts: 59

PostPosted: Mon Sep 26, 2005 8:28 pm    Post subject: Grow up already Reply with quote

*yawn*

Are we still on this? Look, if you're going to run Gentoo on production servers, use it as the metadistribution it is, and do your own release engineering.

If you want a stable release for which updates guarantee backward-compatibility for X years (whereafter you're fscked for support), buy something like RedHat or SuSE.

Gentoo provides something else. Something quite exciting and revolutionary: the option of a rolling metadistribution that never reaches "end of life", from which to cut your own releases, service packs, emergency upgrade packages, whetever.

But this is something that requires you to take more responsibility for the process of keeping systems current.

If you're going to criticise Gentoo, it's worthwhile taking the time to understand what it is. Don't let your mostly painless experience of blind upgrades lull you into a false sense of security. Understand what you're dealing with and behave accordingly.

This stuff is for children. It's also for servers. It's just not for children operating servers.
Back to top
View user's profile Send private message
Display posts from previous:   
This topic is locked: you cannot edit posts or make replies.    Gentoo Forums Forum Index Gentoo Chat All times are GMT
Goto page Previous  1, 2, 3, ... 9, 10, 11  Next
Page 2 of 11

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum