[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [cobalt-users] moving users from group to another group



Julius wrote:

> Jeff L. wrote:
> 
> > The postgreSQL database is bad; I thought we'd agreed on that a
> > long time ago;
> 
> Do you mean it is something which will always be bad
> in your opinion, or in this case has gone bad?
> I was never saying it wasn't, I know it is,
> but it isn't as bad as you think it is, it only
> has a couple of bad entries that need to be rebuilt
> along the way.

My recollection is you made a post showing it (or parts of it) were
empty when you looked.  Have you done meta-verify on it?  Do you know
it's good?

> If you would know my situation, you would not be restoring
> either. It simply is not an option.

Okay, I said you'd decided against the option.

> The RaQ is doing
> dedicated webmail and email mostly. Why it is not serving
> many websites is besides the point, it can be used this
> way and it was the best option in this particular case.
> The amount of webmail and email users is that large,
> a Qube (for example) wouldn't cope with it that easy.
> *they* decided to buy a RaQ4i, after my advice to do so.

You're using a Raq in an environment it really isn't designed for;
that's okay, but it may cause you "issues" downstream.

> First; I can't switch off the admserv, users need to be able
> to change their passwords by use of the GUI, which works
> nicely right now.

Okay.

> The RaQ4 is in 'production' as we speak,
> restoring it would mean a lot of things; The failoverserver
> to work en be there (right now it isn't, thanks to KPNQwest),

Fixable, I presume.

> to have all user data (user prefs/mail etc.) backed up
> the way it is now and be put back the way it is now, which
> can't be done;

Actually pretty easy; linux system admins do this sort of work, both
planned, and unplanned under pressure, all the time.

> The GUI stopped functioning BEFORE anybody
> knew, so a lot of things have been done with somehow
> non-functional postgres and sendmail databases.

So postgreSQL data is bad?

> Putting those accounts, passwords, and aliases back
> the way they are now would be an impossible task,
> and require the exact same misery to go through
> where I'm going through right now.

Not really; you can save backups of the files they're in, and manually
restore them later, then run meta-verify to restore the PostgreSQL
database.

> So the only option is to re-create the postgres databases
> in a way that the GUi will show the default site again.

Okay...

> > you've spent more time than that by now.
> 
> No I haven't. You just don't know the details involved.
> Doing a restore would imply exactly what I'm doing right now.

But you'd only have to do it once and then the system would work.  If
you don't give us enough details then of course we can't suggest
anything to help you.

> >> since I've done a quotaoff for /dev/hda4 and
> >> cobalt's postgresql need at least some quotum,
> >> it can't work without one.
> >
> > I don't think it's working now.
> 
> Well I'm still getting quota warnings over email
> from the machine. Also, when creating a user using
> (c)adduser, I can;t because the user quota is reset
> the wrong way for the default user/site settings.

I thought we went over this before; I still think it's because the
quotas are set to 0 because of an error in postgreSQL data.

> > Do you want to turn off the Cobalt gui entirely and run the box
> > completely manually?  If you do, let me know; I have some notes on
> that somewhere.  But if so you won't be able to use any of the
> > Cobalt tools; not even the cobalt/adduser.
> 
> Like I wrote: It would be better if Sun Cobalt had created
> a graphic interface for the shell, to function as a
> complete backup for the GUI.

And if wishes were fishes ....

I thought we were talking about reality, not about what-ifs.  Sorry for
the misunderstanding.  So this isn't a request for help?  It's a rant?

> >> the main trouble is; editing site quota/settings
> >> through use of the commandline seems impossible.
> >
> > Why?  Because the Cobalt system is somehow overriding them?  You
> can fix that by deleting the Cobalt gui, if you really don't want
> to use it.
> 
> No, I would need to delete the postgres databases,

No, you wouldn't.

> DNS errors are out, that I know in this case.
> Note that I know this exact same erratic behavior
> on RaQs WITHOUT ever having touched the shell.

I'm considered by some to be an expert on DNS; I've even written a book
on it (manuscript not yet accepted by any publisher).  Yet I made two
"dumb" DNS mistakes in the last few months that some knowledgeable
people on this and other lists helped me with.  It can happen.

> There's a lot of talk about the 'bug' on the raqfaq pages.
> It has to do with the order in which you enter domains
> and their aliases etc. This, in my opinion, is simply
> faulty design by Cobalt.

I don't know what pages you're talking about, but if you're talking
about how the alias files and/or the virtusertable file are set up I
know for a fact (both from my own exhaustive tests over the years, and
from documentation in the Sendmail "bat" book and the Sybex Linux
Sendmail Administration book) that the order in either virtusertable and
in aliases don't matter, since both are compiled into database files
before they're used.  As I've said, you can trace email through
sendmail; I strongly urge you get one of the books I mentioned. 
Subscribing to a good sendmail mailing list might be helpful as well.

> Why are you so bound on people needing to 'learn' things
> before they can solve their problems? It is totally
> unnecessary, especially in this case, to 'know what
> you're doing with sendmail'.

Really?

> I don't even have to
> configure sendmail in any other way than it has been
> thus far. It might be better for Sun to find a way
> to create an update pkg for sendmail, since we're
> behind a long way when using a RaQ with the
> version of sendmail currently on it.

Sun isn't going to do that.  So live with it.

> > As long as you realize it won't any longer be running as a RaQ,
> > but rather as some kind of hybrid,
> 
> Why would it be? It hasn't changed a thing if I'm back
> at what it was like and have the GUI working again.

You didn't say anything about getting the gui again; your post said
"that" issue; referring to the issue immediately above, I presumed you
meant the sendmail problem.

> > I know several people who've successfully run RaQs without a
> > virtual site,
> 
> well, go check the admin site for a RaQ4i, create NO
> virtual site and then hope it will receive email for
> more than one domain name on that default site,
> by only using the GUI. I wish you luck.

What do you mean by receive email for more than one domain name?  Do you
mean the same addresses but with other domain names?  That should work
fine (just go to Site Settings and Email Server Aliases, and set up DNS
properly).  It works fine now on the RaQ4 I just tested it on, but the
RaQ already has multiple virtual sites set up on it, so according to
your post, it's not a good test.  I'll have a clean RaQ4i here in a few
days; I'll test it then without any virtual sites.

If that's not what you mean, then what do you mean?  There are only two
ways to set up a RaQ to receive email for multiple domains that I can
think of; one is to use virtual sites, and the other is to do it with
manual changes to the virtusertable file and a few other sendmail files.

I've done it over and over again both ways, and it works both ways, so I
don't know what you mean.

> > and there's no reason why the RaQ "sauce" requires one.
> 
> Well I've discovered that it does.
> Try to add users without there being a virtual domain name.

Been there, done that.  So have clients who rent RaQs from us.  Not
since the latest updates, though, so there could be something to what
you say.  I'll try it again next week, when I have a virgin RaQ4.

> > I don't know what
> > originally caused your problem, and I understand you cannot
> > use it now, but that doesn't mean it can't be used by anyone else.
> 
> >From what I've read the past years, I'm not the only one having
> problems with the default site next to virtual site config
> on a Cobalt RaQ.

What do you mean by default site next to virtual site?

But please leave the rants out of your replies; make it a bit easier for
us to help you.

Jeff
-- 
Jeff Lasman <jblists@xxxxxxxxxxxxx>
Linux and Cobalt/Sun/RaQ Consulting
nobaloney.net
P. O. Box 52672, Riverside, CA  92517
voice: (909) 778-9980  *  fax: (702) 548-9484