[Dovecot] Upgrade to 2.0 - Not so smooth
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull
Just some real time feedback. I don't know what dovenull user is or why it is necessary.
On 15.8.2010, at 17.03, Marc Perkel wrote:
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull
Just some real time feedback. I don't know what dovenull user is or why it is necessary.
I added now to http://wiki2.dovecot.org/Upgrading/2.0 :
- Dovecot uses two internal users now by default: dovenull and dovecot. You need to create the dovenull user or change default_login_user setting.
On 8/15/2010 9:10 AM, Timo Sirainen wrote:
On 15.8.2010, at 17.03, Marc Perkel wrote:
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull
Just some real time feedback. I don't know what dovenull user is or why it is necessary. I added now to http://wiki2.dovecot.org/Upgrading/2.0 :
- Dovecot uses two internal users now by default: dovenull and dovecot. You need to create the dovenull user or change default_login_user setting.
might want to automatically add dovenull to the install script if possible. Perhaps a better error message than the one above that indicates the solution is to create the dovenull user.
Everything seems to be working. Is there a "what's new" feature list anywhere? Anything cool I might want to try?
On 2010-08-15 09:44:49 -0700, Marc Perkel wrote:
might want to automatically add dovenull to the install script if possible. Perhaps a better error message than the one above that indicates the solution is to create the dovenull user.
I dont think you can know all the different tools distros use to create users. (adduser vs useradd e.g.) so documenting this is the better approach.
darix
-- openSUSE - SUSE Linux is my linux openSUSE is good for you www.opensuse.org
On 8/15/2010 9:58 AM, Marcus Rueckert wrote:
On 2010-08-15 09:44:49 -0700, Marc Perkel wrote:
might want to automatically add dovenull to the install script if possible. Perhaps a better error message than the one above that indicates the solution is to create the dovenull user. I dont think you can know all the different tools distros use to create users. (adduser vs useradd e.g.) so documenting this is the better approach.
darix
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
On 2010-08-15 14:01:25 -0700, Marc Perkel wrote:
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
that is 2 out of many solutions. and depending of the admin he might create the user in an ldap tree or a nis server.
also packager dont really like users created from the buildsystem.
darix
-- openSUSE - SUSE Linux is my linux openSUSE is good for you www.opensuse.org
On 8/15/2010 2:32 PM, Marcus Rueckert wrote:
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
On 2010-08-15 14:01:25 -0700, Marc Perkel wrote: that is 2 out of many solutions. and depending of the admin he might create the user in an ldap tree or a nis server.
also packager dont really like users created from the buildsystem.
darix
Perhaps a better error message that says something like:
Dovecot needs a user named "dovenull" to work. Please create user "dovenull" and try again.
On 8/15/2010 5:49 PM, Marc Perkel wrote:
On 8/15/2010 2:32 PM, Marcus Rueckert wrote:
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
On 2010-08-15 14:01:25 -0700, Marc Perkel wrote: that is 2 out of many solutions. and depending of the admin he might create the user in an ldap tree or a nis server.
also packager dont really like users created from the buildsystem.
darix
Perhaps a better error message that says something like:
Dovecot needs a user named "dovenull" to work. Please create user "dovenull" and try again.
I agree its not as simple as people would think but, if dovecot would log everything relevant, dovecot could use its own logs to know what environment it is within.
Jerrale G. SC Senior Admin
On 15/08/2010 22:49, Marc Perkel wrote:
On 8/15/2010 2:32 PM, Marcus Rueckert wrote:
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
On 2010-08-15 14:01:25 -0700, Marc Perkel wrote: that is 2 out of many solutions. and depending of the admin he might create the user in an ldap tree or a nis server.
also packager dont really like users created from the buildsystem.
darix
Perhaps a better error message that says something like:
Dovecot needs a user named "dovenull" to work. Please create user "dovenull" and try again.
No.
The first sentence of your proposed message is FALSE.
On 15/08/2010 17:10, Timo Sirainen wrote:
- Dovecot uses two internal users now by default: dovenull and dovecot. You need to create the dovenull user or change default_login_user setting.
Dovecot does NOT need a user named "dovenull" to work.
The second sentence is predicated on the first and as such also falls down.
Dovecot needs a user named by the default_login_user setting in order to work. Whilst this defaults to "dovenull", there is nothing to stop you setting it to something else (e.g. "flibble").
The point is that the answer is more of a commentary and the right solution depends on the situation.
Rather than try to squeeze solution summaries into every error message, I would suggest having a link (perhaps into the Dovecot wiki) in the error message, e.g.
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull (see http://wiki2.dovecot.org/Errors/NoDefaultLoginUser)
(Note that the above URL is hypothetical.)
Then have a commentary and suggested solutions in the wiki.
Alternatively, the link could rooted at a notional documentation root like:
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull (see documentation "Errors/NoDefaultLoginUser")
so that users could refer to a local copy of the documentation rather than encouraging load on the Dovecot wiki server(s).
Bill
On 8/16/2010 2:45 AM, William Blunn wrote:
On 15/08/2010 22:49, Marc Perkel wrote:
On 8/15/2010 2:32 PM, Marcus Rueckert wrote:
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
On 2010-08-15 14:01:25 -0700, Marc Perkel wrote: that is 2 out of many solutions. and depending of the admin he might create the user in an ldap tree or a nis server.
also packager dont really like users created from the buildsystem.
darix
Perhaps a better error message that says something like:
Dovecot needs a user named "dovenull" to work. Please create user "dovenull" and try again.
No.
The first sentence of your proposed message is FALSE.
On 15/08/2010 17:10, Timo Sirainen wrote:
- Dovecot uses two internal users now by default: dovenull and dovecot. You need to create the dovenull user or change default_login_user setting.
Dovecot does NOT need a user named "dovenull" to work.
The second sentence is predicated on the first and as such also falls down.
Dovecot needs a user named by the default_login_user setting in order to work. Whilst this defaults to "dovenull", there is nothing to stop you setting it to something else (e.g. "flibble").
The point is that the answer is more of a commentary and the right solution depends on the situation.
Rather than try to squeeze solution summaries into every error message, I would suggest having a link (perhaps into the Dovecot wiki) in the error message, e.g.
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull (see http://wiki2.dovecot.org/Errors/NoDefaultLoginUser)
(Note that the above URL is hypothetical.)
Then have a commentary and suggested solutions in the wiki.
Alternatively, the link could rooted at a notional documentation root like:
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull (see documentation "Errors/NoDefaultLoginUser")
so that users could refer to a local copy of the documentation rather than encouraging load on the Dovecot wiki server(s).
Bill
Maybe dovecot could use the "nobody" user if dovenull doesn't exist?
On 2010-08-16 07:16:46 -0700, Marc Perkel wrote:
Maybe dovecot could use the "nobody" user if dovenull doesn't exist?
you could just read the warning message, create the user and stop worrying?
and no ... it shouldnt fallback to nobody. a broken set up is a broken setup and needs to be fixed.
darix
-- openSUSE - SUSE Linux is my linux openSUSE is good for you www.opensuse.org
On Mon, 2010-08-16 at 07:16 -0700, Marc Perkel wrote:
Maybe dovecot could use the "nobody" user if dovenull doesn't exist?
The entire point of dovenull user is that it's not used by anything else. The only things I'm going to change here is to make the error message somewhat better for v2.0.1 and add a note about this user to NEWS file / release announcement.
On Aug 15, 2010, at 4:01 PM, Marc Perkel wrote:
Seems to me that it wouldn't take a lot of code to at lease find if they have adduser or useradd.
Or in the case of OS X, dscl...
Documenting this is good, but perhaps the configure script could check if the user exists, and instruct the user to create this manually before going any further?
On Aug 15, 2010, at 9:04 PM, Jim wrote:
On Aug 15, 2010, at 4:01 PM, Marc Perkel wrote:
Seems to me that it wouldn't take a lot of code to at lease find if
they have adduser or useradd.Or in the case of OS X, dscl...
Documenting this is good, but perhaps the configure script could
check if the user exists, and instruct the user to create this
manually before going any further?
And some may be building binaries packages for distribution so forcing
user creation on a build server would be extra unhelpful work.
On thing I was considering was if patching to allow configure to set --
default_login_user=_dovenull and --default_internal_user=_dovecot. I
don't think this is a big deal but it would allow package builders to
build packages with default names that match their distributions.
BTW, things are fine the way they are. The fewer options that meet the
needs the better.
Regards, Bradley Giesbrecht
On 2010-08-15 12:03 PM, Marc Perkel marc@perkel.com wrote:
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull
Just some real time feedback. I don't know what dovenull user is or why it is necessary.
It would be a lot smoother (and this probably applies to all other software you use) if you would spend at least 5 SECONDS doing a little research before wasting everyone's (especially Timo's) time with questions that you can easily answer yourself:
http://www.lmgtfy.com/?q=dovenull
For dovecot stuff, I've found that simply adding the word 'wiki' after the other search term will almost always provide the link to the wiki page describing whatever it is:
http://www.lmgtfy.com/?q=dovenull+wiki
Yes, this is a pet peeve of mine...
--
Best regards,
Charles
On 8/16/2010 4:11 AM, Charles Marcus wrote:
On 2010-08-15 12:03 PM, Marc Perkelmarc@perkel.com wrote:
Starting Dovecot Imap: Fatal: service(pop3-login) User doesn't exist: dovenull
Just some real time feedback. I don't know what dovenull user is or why it is necessary. It would be a lot smoother (and this probably applies to all other software you use) if you would spend at least 5 SECONDS doing a little research before wasting everyone's (especially Timo's) time with questions that you can easily answer yourself:
http://www.lmgtfy.com/?q=dovenull
For dovecot stuff, I've found that simply adding the word 'wiki' after the other search term will almost always provide the link to the wiki page describing whatever it is:
http://www.lmgtfy.com/?q=dovenull+wiki
Yes, this is a pet peeve of mine...
Timo's software standards, and mine, are higher than the average open source project. When an install id done right then you don't have to go to the wiki for anything. You run it and it just works. That's why people pay for Windows and Macs and more people use it than Linux because it just works. You start an upgrade anf click NEXT AGREE NEXT NEXT NEXT FINISH and everything just works. That's the way Linux should be.
So - even though something might be a minor detail, when you get the minor details right then you get software the "it just works" which is in my opinion the highest thing one can say about a program. And it's who dovecot is so popular.
On Mon, Aug 16, 2010 at 07:23:50AM -0700, Marc Perkel wrote:
Timo's software standards, and mine, are higher than the average open
source project. When an install id done right then you don't have to go
to the wiki for anything. You run it and it just works. That's why
people pay for Windows and Macs and more people use it than Linux
because it just works. You start an upgrade anf click NEXT AGREE NEXT
NEXT NEXT FINISH and everything just works. That's the way Linux should be.So - even though something might be a minor detail, when you get the
minor details right then you get software the "it just works" which is
in my opinion the highest thing one can say about a program. And it's
who dovecot is so popular.
So are you saying a server software which brings a lot of new features to cope with complex environments shall tweak itself by magic, as any good software should upgrade without user intervention?
You wouldn't even need documentation as the new features would be self-explanatory?
*NO FRIGGIN' WAY*
On 8/16/2010 7:37 AM, Thomas Leuxner wrote:
On Mon, Aug 16, 2010 at 07:23:50AM -0700, Marc Perkel wrote:
Timo's software standards, and mine, are higher than the average open source project. When an install id done right then you don't have to go to the wiki for anything. You run it and it just works. That's why people pay for Windows and Macs and more people use it than Linux because it just works. You start an upgrade anf click NEXT AGREE NEXT NEXT NEXT FINISH and everything just works. That's the way Linux should be.
So - even though something might be a minor detail, when you get the minor details right then you get software the "it just works" which is in my opinion the highest thing one can say about a program. And it's who dovecot is so popular. So are you saying a server software which brings a lot of new features to cope with complex environments shall tweak itself by magic, as any good software should upgrade without user intervention?
You wouldn't even need documentation as the new features would be self-explanatory?
*NO FRIGGIN' WAY*
No - I'm saying that an upgrade that does exactly the same thing as the earlier version should "just work" without having to research cryptic error messages you get after the new software fails to load. What I'm saying is that Linux should be as easy as Windows.
Am 16.08.2010 um 19:22 schrieb Marc Perkel:
No - I'm saying that an upgrade that does exactly the same thing as the earlier version should "just work" without having to research cryptic error messages you get after the new software fails to load. What I'm saying is that Linux should be as easy as Windows.
And that's the point. It does not do things like the earlier version as privileges have been removed from processes. All of that has been noted on the list during development and in the documentation. There was even a poll around the actual name to be used.
On Mon, 2010-08-16 at 10:22 -0700, Marc Perkel wrote:
a
No - I'm saying that an upgrade that does exactly the same thing as the earlier version should "just work" without having to research cryptic error messages you get after the new software fails to load. What I'm saying is that Linux should be as easy as Windows.
A new MAJOR version release should ring bells that you need to read for any possible problems. This is true for other widely popular software, ie: apache.
I will admit, certain distro vendors such as debian/RH/Fedora/Ubuntu , do, do much software a dis-service by their hacks and non standard file locations, because people with problems come to us for support expecting us to fix a mess created by someone else. That I know annoys a lot of people trying to convert to linux from windows and you're right, then they say bugger it and go back to windows. but.. everyone wants to be special I suppose :)
On Tue, Aug 17, 2010 at 08:19:03AM +1000, Noel Butler wrote:
I will admit, certain distro vendors such as debian/RH/Fedora/Ubuntu , do, do much software a dis-service by their hacks and non standard file locations,
That depends on what you define "standard". Why do so many software authors ship a default configuration which violates posix and fhs?
Greetings Marc
--
Marc Haber | "I don't trust Computers. They | Mailadresse im Header Mannheim, Germany | lose things." Winona Ryder | Fon: *49 621 72739834 Nordisch by Nature | How to make an American Quilt | Fax: *49 3221 2323190
On Wed, 18 Aug 2010 12:58:38 +0200 Marc Haber mh+dovecot@zugschlus.de articulated:
That depends on what you define "standard". Why do so many software authors ship a default configuration which violates posix and fhs?
Probably due to the fact that while many *.nix/BSD OSs claim to support 'posix' etc. they don't fully do so. The problems involved in getting certain program to operate between *.nix and BSD alone is enough to make a grown man cry. Vendors are simply doing what they have to in order to insure that their applications will work on as wide a scale as possible. The Filesystem Hierarchy Standard (FHS) is not even standard between various *.nix and BSD systems, let alone a Microsoft one. It seems that everyone has to reinvent the wheel.
-- Jerry ✌ Dovecot.user@seibercom.net
Disclaimer: off-list followups get on-list replies or get ignored. Please do not ignore the Reply-To header.
Why is it called a funny bone when it hurts so much?
Marc Perkel put forth on 8/16/2010 12:22 PM:
No - I'm saying that an upgrade that does exactly the same thing as the earlier version should "just work" without having to research cryptic error messages you get after the new software fails to load. What I'm saying is that Linux should be as easy as Windows.
Would you like Linux based operating systems/applications to be as insecure as Windows?
Would you like Linux based operating systems/application log entries to be as worthless when attempting to troubleshoot something as with Windows?
Would you like Linux based operating systems/applications to change the location of menu items and configuration options with each upgrade just for the sake of "change"? So people don't think "what the hell did I just pay $500 for? Nothing changed!?"
Etc, etc, etc.
Everything is a trade-off Marc. All the effort that Microsoft puts into making things "easy" takes resources and focus away from other areas, often critical areas. Those other areas are more critical for Linux/Unix systems and applications because people need reliability from them more than they need ease of installation.
The Linux world doesn't do everything right, and the MS world doesn't do everything wrong. But overall I think the Linux world tends to strike a better overall balance. It all comes down to expectations. You _expect_ MS things to work a certain way. The Linux world is inherently different. So you shouldn't automatically expect things in the Linux world to work "The Microsoft Way". And frankly you shouldn't want it that way either.
-- Stan
Em 16/08/2010 11:23, Marc Perkel escreveu:
Timo's software standards, and mine, are higher than the average open source project. When an install id done right then you don't have to go to the wiki for anything. You run it and it just works. That's why people pay for Windows and Macs and more people use it than Linux because it just works. You start an upgrade anf click NEXT AGREE NEXT NEXT NEXT FINISH and everything just works. That's the way Linux should be.
So - even though something might be a minor detail, when you get the minor details right then you get software the "it just works" which is in my opinion the highest thing one can say about a program. And it's who dovecot is so popular.
hey Marc, you're not alone !!! As well as you, i also expect
software updates to be always perfect and magic .... so i dont have to have a clue of what i'm real doing, as softwares will take care of all the inteligence that i should have.
but, i think different from you, i understand and accept that
there's no magic.
upgrades on complex environment MUST be planned, new version
features should be understood and in almost all cases, including some M$ updates of complex software/ environment, lots of pre-upgrade and post-upgrades actions should be taken.
i would love to just click yes yes and everything works as magic
.... but, unfortunelly, i accept that wont happen in the real (and most all the times complex) scenarios we have.
--
Atenciosamente / Sincerily,
Leonardo Rodrigues
Solutti Tecnologia
http://www.solutti.com.br
Minha armadilha de SPAM, NÃO mandem email
gertrudes@solutti.com.br
My SPAMTRAP, do not email it
On Mon, 2010-08-16 at 11:42 -0300, Leonardo Rodrigues wrote:
hey Marc, you're not alone !!! As well as you, i also expect
software updates to be always perfect and magic .... so i dont have to have a clue of what i'm real doing, as softwares will take care of all the inteligence that i should have.
Yes, that's what I expect distribution packages to do. But upstream packages can't do that. There are just way too many different ways people use them.
On 2010-08-16 10:23 AM, Marc Perkel marc@perkel.com wrote:
That's why people pay for Windows and Macs and more people use it than Linux because it just works. You start an upgrade anf click NEXT AGREE NEXT NEXT NEXT FINISH and everything just works. That's the way Linux should be.
You're not serious?? I can't count how many times an update of Windows or a windows program didn't go as expected or planned, but this has absolutely nothing to do with the fact that your question was posted to this list because you were too lazy to even make a half-hearted attempt to figure out the problem yourself.
So - even though something might be a minor detail, when you get the minor details right then you get software the "it just works" which is in my opinion the highest thing one can say about a program. And it's who dovecot is so popular.
Irrelevant. Sure, I'd love for it to just work like magic every time, but reality is very different, and if you are going to be too lazy to try to fix something yourself, you should expect the kind of responses you have gotten.
--
Best regards,
Charles
participants (14)
-
Bradley Giesbrecht
-
Charles Marcus
-
Jerrale G
-
Jerry
-
Jim
-
Leonardo Rodrigues
-
Marc Haber
-
Marc Perkel
-
Marcus Rueckert
-
Noel Butler
-
Stan Hoeppner
-
Thomas Leuxner
-
Timo Sirainen
-
William Blunn