Mailing List Archive

On 2007-12-20, at 0849, Aaron Goldblatt wrote:
>
> Financial considerations have forced me to undertake something no
> admin ever wants to do ... a server migration. But the money I'm
> saving on a monthly basis is going to have to make it worth it
> (especially since I'm getting a better machine).
>
> I'd like the wizards here to please take a look at my plan and point
> out where I've screwed it up or what I've forgotten.

not so much "missing", but something which might make your life easier.

if you're using vpopmail, i wrote a script called "migrate-domain"
which builds tarballs of each mailbox, autoresponder, and ezmlm list
directory, and generates a shell script which re-creates the same
items on another server, using the vpopmail command line tools. this
gets around the issues of domain and mailbox hashing and the "vpasswd"
file not having the correct directory in it.

and if you're using something else, feel free to poke through the
script and use it for ideas.

http://qmail.jms1.net/scripts/migrate-domain.shtml

----------------------------------------------------------------
| John M. Simpson --- KG4ZOW --- Programmer At Large |
| http://www.jms1.net/ <jms1@jms1.net> |
----------------------------------------------------------------
| http://video.google.com/videoplay?docid=-1656880303867390173 |
----------------------------------------------------------------

_____

From: Aaron Goldblatt [mailto:qmailconfused@gmail.com]
To: qmail@list.cr.yp.to
Sent: Thu, 20 Dec 2007 04:49:01 -0900
Subject: Server Migration

Financial considerations have forced me to undertake something no
admin ever wants to do ... a server migration. But the money I'm
saving on a monthly basis is going to have to make it worth it
(especially since I'm getting a better machine).

I'd like the wizards here to please take a look at my plan and point
out where I've screwed it up or what I've forgotten.

I'm coming FROM:
- netmail 1.05/LWQ + lukasfeiler.com-bigqmail-20070526
- djbdns

I'm going TO:
- netqmail 1.06/LWQ + lukasfeiler.com-bigqmail-20070526
- djbdns


So the plan is ...

1) Create necessary user accounts on new box
2) Configure new box for rcpthosts/locals/virtualdomains to mirror old box
3) Create necessary delivery instructions on new box (.qmail-*, etc)
4) Modify djbdns/root/data to reflect new server as primary MX and old
server as backup MX
5) Remove domains from locals/virtualdomains on old box, but leave in rcpthosts
6) Restart qmail on old box, and immediately after ...
7) Make djbdns/root/data to make new MX settings live
8) Wait out a reasonable propogation delay
9) Modify djbdns to remove backup MX
10) Decommission old server

Feedback muchly appreciated. Thanks!

ag
Did you figure things out or get any replies ?

-Dee

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On Monday, April 8 at 01:51 AM, quoth Aaron Goldblatt:
> so, any directions on where i can look or how i can trace this
> further would be most helpful. i'm obviously missing something, but
> for the life of me do not see what.

I believe the CDB database format is not fully cross-platform, and if
your old computer is 32-bit and your new one is 64-bit, you may need
to simply rebuild the validrcptto.cdb database.

~Kyle
- --
The purpose of computing is insight, not numbers.
-- Richard W. Hamming
-----BEGIN PGP SIGNATURE-----
Comment: Thank you for using encryption!
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=Mzb1
-----END PGP SIGNATURE-----

On Mon, Apr 8, 2013 at 11:01 AM, Kyle Wheeler <kyle-qmail@memoryhole.net> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----

> I believe the CDB database format is not fully cross-platform

Really? I'm convinced that the database format is fixed,
cross-platform, etc. etc.

quote> Positions, lengths, and hash values are 32-bit quantities, stored in
quote> little-endian form in 4 bytes.

I think you should look elsewhere.

- M

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On Monday, April 8 at 11:12 AM, quoth Michael Sierchio:
> On Mon, Apr 8, 2013 at 11:01 AM, Kyle Wheeler <kyle-qmail@memoryhole.net> wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>
>> I believe the CDB database format is not fully cross-platform
>
> Really? I'm convinced that the database format is fixed,
> cross-platform, etc. etc.
>
> quote> Positions, lengths, and hash values are 32-bit quantities, stored in
> quote> little-endian form in 4 bytes.
>
> I think you should look elsewhere.

AAAAnd if you actually read the *code* of DJB's CDB software, it
defines "32-bit unsigned integer" as "unsigned int" (see uint32.h),
which on machines that use the ILP64 data model, is actually a 64-bit
integer. This is a consequence of using software that pre-dates the
establishment of the uint32_t datatype: it's darn hard to create an
integer of a specific (large-ish) size in a portable way. You can
consider that a bug in his software if you like; it certainly means it
can violate the CDB spec on some machines with some compilers.

But, whether this is the issue or not, it can't hurt to give my
suggestion a shot.

~Kyle
- --
The important work of moving the world forward does not wait to be
done by perfect men.
-- George Eliot
-----BEGIN PGP SIGNATURE-----
Comment: Thank you for using encryption!
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=806P
-----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On Monday, April 8 at 01:51 AM, quoth Aaron Goldblatt:
>OBSERVED BEHAVIOR
>on preg, *@goldblatt.net comes in and is bounced immediately with "550
>sorry, no mailbox here by that name. (#5.1.1)" including stuff to
>known-good addresses that should be passed to google.

When you say "bounced immediately", do you mean it's *rejected* during
the SMTP session (i.e. you are running into a validrcptto problem), or
is it accepted and a bounce message is generated (i.e. you are running
into a delivery problem)?

~Kyle
- --
Only the fool hopes to repeat an experience; the wise man knows that
every experience is to be viewed as a blessing.
-- Henry Miller
-----BEGIN PGP SIGNATURE-----
Comment: Thank you for using encryption!
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=jAVZ
-----END PGP SIGNATURE-----

On 08/04/2013 16:37, Kyle Wheeler wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> On Monday, April 8 at 01:51 AM, quoth Aaron Goldblatt:
>> OBSERVED BEHAVIOR
>> on preg, *@goldblatt.net comes in and is bounced immediately with "550
>> sorry, no mailbox here by that name. (#5.1.1)" including stuff to
>> known-good addresses that should be passed to google.
> When you say "bounced immediately", do you mean it's *rejected* during
> the SMTP session (i.e. you are running into a validrcptto problem), or
> is it accepted and a bounce message is generated (i.e. you are running
> into a delivery problem)?
>
> ~Kyle
> - --
> Only the fool hopes to repeat an experience; the wise man knows that
>
The issue is his validrcptto.cdb
I think he should run his test without the wildcard email address

That is the standard bounce for rejection during smtp session ... it is
not a delivery issue

From what I can make up his old box was not rejecting the invalid
addresses simply deleting the mails
That is not the behaviour of my validrcptto patch.

Sending from a cell phone. Apologies.

> When you say "bounced immediately", do you mean it's *rejected* during
> the SMTP session (i.e. you are running into a validrcptto problem), or
> is it accepted and a bounce message is generated (i.e. you are running

The error message is geneated by Google (sending from an unrelated gmail
account). If you think regenerating the cdb would help i can, but both
systems are 64 bit debian, so i doubt endianness is the culprit. But i am
willing to try and report this evening.

Ag