Re: Migration murder from 3.2 to 3.10 (Debian 11 to 13) – problems with shared folders

Am 29.04.26 um 09:40 schrieb Frank Richter:
> Quoting Jean Charles Del=C3=A9pine:
>>
>> Quoting Frank Richter <[email protected]>:
>>
>>> We=E2=80=99re facing a problem with shared folders. Our shared folders =
have this=20
>>> scheme: "group.sub.name"
>>> The problem occurs when creating a new shared folder=20
>>> "group.sub.othername" on backend "packet":
>>>
>>> 2026-04-28T12:57:40.824198+02:00 packet cyrus/imaps[2647805]: MUPDATE:=
=20
>>> can't commit mailbox entry for 'group'
>>> 2026-04-28T12:57:40.828960+02:00 packet cyrus/imaps[2647805]: Fatal=20
>>> error: Internal error: assertion failed: imap/imapd.c: 1494:=20
>>> !open_mailboxes_exist()
>>>
>>> Hm, "group" shouldn=E2=80=99t be a mailbox, at least it wasn't on cyrus=
-imapd=20
>>> 3.2.11. But now there is :
>>
>> In the Cyrus IMAP 3.6 Releases you will find :
>>
>> Major changes since the 3.4 series
>>
>> =C2=A0=C2=A0=C2=A0 All members of a newly created mailbox hierarchy will=
 now be=20
>> selectable mailboxes. In other words, when creating mailbox=20
>> user.foo.A.B.C, both user.foo.A and user.foo.A.B will also be created as=
=20
>> real mailboxes.=20
> I read this =E2=80=A6 and hoped that this is the case only for user.* mai=
lboxes.
> Now in my setup this means that I can=E2=80=99t distribute my shared fold=
ers=20
> group.* between backends any longer =E2=80=93 all group mailboxes have to=
 be on=20
> the same backend server now. This hurts (a bit). If this would be the sam=
e=20
> for user mailboxes, then a murder setup makes nearly no sense.=20


How does other admins of a Cyrus murder handle this, is anyone using shared=
=20
folders? We use it a lot for group scenarios, such as mailboxes for project=
,=20
departments etc. So for us the "pinning" of all group folders to only one=20
backend server isn't optimal.

Would it be reasonable to have a configuration option for the namespace of=
=20
shared mailboxes, so that for this name no mailbox will be created?

Thanks,
Frank

--=20
Frank Richter, Chemnitz University of Technology, Germany


------------------------------------------
Cyrus: Info
Permalink: https://cyrus.topicbox.com/groups/info/Te7a1ee0606b2df5d-Mf91dfa=
e61179dff984bd124a
Delivery options: https://cyrus.topicbox.com/groups/info/subscription