Topic: test upgrade 10.1 to 11 beta

[logs]

I tested the upgrade with MigrateHelper from a smaller server with 100 users from 10.1 to 11 Beta.
The upgrade apparently worked well. server-manager login OK
I performed a dnf update and installed Contribs:

smeserver-certificate
smeserver-learn
smeserver-migratehelper
smeserver-service_control
smeserver-systemd-control
smeserver-vacation
smeserver-wbl


I tried logging in as a RoundCube users ok,
Install smeserver-qemu-guest-agent <<<- Fail deps

After install smeserver-qmHandle fail all logins on roundcube and not read queue on server-manager.
After remove smeserver-qmHandle logins ok

Tested on virtual machine proxmox:
Host 32 x AMD EPYC 7302P 16-Core Processor (1 Zócalo)
/ - /boot - swap ---> virtio0
/home/e-smith/files ---> virtio1

screenshots of hardware and options

Any suggestions on testing?

[ReetP]

As previously mentioned, most contribs have still to be worked on.

Better to test one at a time, check bugs and src.koozali.org to see what has been worked on etc. Otherwise you won't be able to see the wood for the trees.

If you DM me with your full name & a email address for notifications I can add you to Rocket.Chat so you can help with specific testing and keep up with development.

[ReetP]

Estas en España??

Vivo cerca Valencia pero soy Ingles

[logs]

Name = Jose Maria Garcia Sanchez
Mail - "Not public"

[logs]

 Si en Huesca y Zaragoza, español.

[logs]

I will proceed to test an upgrade of a server with 700 users, using a similar procedure
but changing to different types of processors and activacion vIOMMU and NUMA.
q35/host
x86-64-v2-AES
x86-64-v3
x86-64-v4

[ReetP]

Si en Huesca y Zaragoza, español.

Vale!! Encontado

[ReetP]

Name = Jose Maria Garcia Sanchez
Mail - "Not public"

I have sent your login details. Come and say hello!

[logs]

Upgrade another server
Proxmox 9.0.10
x86-64-v4 does not boot
x86-64-v3 and host, --> boot

Enabling vIOMMU/virtio and NUMA respond more slowly,

I have disabled vIOMMU and IO pressure drops dramatically.
I need to review the case.


Procedure:
Restored production server on another node.
Added vfat small disk to load smeserver.tgz created by migratehelper
Started with sysreccd iso, formatted vfat small disk and transferred smeserver.tgz to vfat disk.
Stop and change the iso to sme-server in the installation, customize the partitioning
vda1 /boot + format
vda2 / + format
vda3 swap
vdb1 /home/e-smit/files (not format contains user data)

On the first reboot, reconfigure the network, remove the vfat disk from /etc/fstab, and add usrquota,grpquota to /home/e-smith/files.
dnf update, reboot, install contribs, and check/disable certain cronjobs.
Only installed contribs:

smeserver-certificate
smeserver-geoip
smeserver-learn
smeserver-migratehelper
smeserver-vacation
smeserver-wbl

I'm reviewing what's in contrib-testing and dev.

I have to see if there is anything that allows me to access user accounts in Roundcube
and bulk export/import users contact books, because that's hell...   

[ReetP]

Regarding your boot issues please ask on Rocket. Lots of us use Proxmox.

That or bug it, but check the Rocky v8 requirements.

Have no idea if any of those contribs work. They may all be broken. Test ONE at a time please and then bug it.

This is still Beta and contribs are slowly being worked on.

Roundcube - no idea as yet. Horde had effectively become unsupported so we had to move on.

[Jean-Philippe Pialasse]

if you manage to write something to import those, we would be happy to integrate it.
As John said we had to make choice with no release from horde for years and php moving on.

[logs]

The migration also doesn't pick up the aliases in Roundcube. The email address for the identities is user@domain.xxx, not the alias.
The set of problems with migrating the email/groupware platform and the remnants of the previous one, in my opinion, don't warrant migrating the production servers.
I think extracting users, passwords, and nicknames and being able to inject them into a clean installation is the appropriate approach, because I see no other solution than restoring a copy of the production servers, changing all passwords to a predefined one, and going user by user, extracting their address books and other groupware applications used, importing them into the clean installation, configuring the identities, etc., and finally applying the exported passwords. Roundcube also doesn't pick up the alias when creating new users.
I'm looking into whether adapting the migratehelper or Lat* script might be the way to go...
In any case, it's a painful operation.

[Jean-Philippe Pialasse]

roundcube needs user to login to create an entry.

issue was with horde previous setting where user preferences where stored in different db entry depending on what  url the user used to log in.  You might end up with multiple profiles for the same horde user.


however looking at your last post i am wondering if you are not in a XY problem.
Would you able to state what your expect and what you observe?
I might save you lote of time rather tha. implementing something first. 

[ReetP]

You do have a Rocket account now.

Much easier discussing there than here.

We don't bite!

[logs]

I was hoping not to have to deploy the servers from scratch.
In all the upgrades, I think since version sme7, I don't even remember, the difficulties were with the address books and little else. Within a period of approximately 7 days, I was able to bring the migrated servers online.

I've never had any services or applications on these servers either, so I understand that the communications platform won't be a problem for most users.

I note that the work you've always done has been brutal and is counted in years, which means that many users are probably still on 9.3. I don't have the knowledge to say whether, with all this work you've done, Horde 6, which is in its alpha version, would have been impossible to attempt.

For my part, this is the end of this thread, and thank you very much for your attention.

[Jean-Philippe Pialasse]

cpanel uses 2 scripts to export horde contact and horde calendar.
while not an in place usage possible it could be used as a start to write one for sme.
It will still miss few other things like tasks. 

- https://hepac-eg.com/FSS-TVH/root/scripts/export_horde_calendars_to_ics
- https://hepac-eg.com/FSS-TVH/root/scripts/export_horde_contacts_to_vcf

[ReetP]

I was hoping not to have to deploy the servers from scratch.

Please be aware that as we have stated multiple times this NOT a policy of Koozali SME but of the upstream distro.

March 2024.

https://forums.koozali.org/index.php/topic,55145.0.html

Your argument lies with RedHat, not us. Don't shoot the messenger.

We have gone to great lengths to try and ameliorate the issues by providing scripts such as migratehelper etc.

We are few and there are limits to what we can achieve.

We can achieve more if people actually help rather than just sit back and complain.

I note that the work you've always done has been brutal and is counted in years

I think this is lost in translation somehow.

We have consistently asked for help for YEARS and get ignored all the time.

How do you expect us to do more when we work as volunteers in our own time with families and mouths to feed?

If knowledgeable experienced people like you stepped up and helped it would all be done a lot faster. We'd love to release fast. We have done a huge amount of unseen backend work over the last year to try and ensure that v12 will be released much sooner. But we can't do the impossible.

which means that many users are probably still on 9.3.

There are lots of people on lots of versions for lots of reasons. Some people also use DOS. Windows 3, XP, 7, 8 etc.

I don't understand them, but that is the way of the world. We can't force them to move. We have no way of contacting the users to ask. For many it is because they don't want to/can't upgrade legacy applications but it is hard to tell.

Again, that is not all down to us. Upgrading Koozali SME is A COLLECTIVE responsibility. If it takes time or is late then EVERYONE is to blame. Not just the few who volunteer to help.

I don't have the knowledge to say whether, with all this work you've done, Horde 6, which is in its alpha version, would have been impossible to attempt.

We looked at Horde carefully and as we have stated, it is essentially unmaintained and we do not have the resources to maintain it ourselves.

https://github.com/horde/webmail/

7 years since the last commit.

It would be way too complex to upgrade and maintain. PHP is advancing far faster than Horde is able to handle. We haven't got the manpower to rewrite Horde 5.x to run on PHP7+

We really had no choice but to change.

You had a opportunity to make your voice heard some time ago. We can't help it if you weren't bothered back then.

https://forums.koozali.org/index.php/topic,55157.0.html

For my part, this is the end of this thread, and thank you very much for your attention.

Let us know if you want to help..... It doesn't fix itself. We still have a mountain of bugs that need verifying.

https://forums.koozali.org/index.php/topic,55312.0.html

However, I assume from your comments that you won't be contributing further. Don't forget - people not contributing is why it takes a long time to build......

I'll close your Rocket account as we are limited on the number of Enterprise licenses we are given and I want to use them for active contributors willing to help us.

If you want to use it to help us then message me soonest.

Thanks.