Nextcloud 14.0.0 Beta 2 has arrived

🎁 Releases
1

Hi,

we just packaged the second beta release for our upcoming Nextcloud 14 version.

You can find as usually on our download server in the prerelease folder:

https://download.nextcloud.com/server/prereleases/nextcloud-14.0.0beta2.zip

We fixed a lot of paper cuts and bugs that were found by all the people that tested the release. It is not perfect yet but issues are being ironed out!

As always: have a backup and know where the docs are to do a rollback (https://docs.nextcloud.com/server/13/admin_manual/maintenance/restore.html for example).

We would ask you for your feedback on this - best is to have it directly on Github - we already added a label for the issues reported especially regarding the Nextcloud 14 beta: “14-beta-feedback” (https://github.com/nextcloud/server/issues?q=is%3Aissue+is%3Aopen+label%3A14-beta-feedback) so we can actually focus to polish the Nextcloud 14 release.

For known bugs and a list of stuff that is scheduled for the beta 3 you can have a look at the overview ticket at https://github.com/nextcloud/server/issues/10502

Happy testing!

Cheers,
–Roeland

2

External users is still broken:

3

Wonderful!

Is this issue planned to be fixed?

1 Like
4

Is known what causes that problem? (I could assist analyzing it. I have a installation with that problem. :wink: )

5

It seems, that it is caused by file chunking. Files are being fully uploaded, but can’t “assebmle” on the storage side.

6

Yes. That’s what happening.

Does anyone know why they are not processed correct after they were uploaded?
Why they are not put togehter?

7

Maybe it has something to do with nginx or some security configurations, but i didn’t find any info anywhere.

8

It may also depends on the OS - I have a problem with NC13.x running in Docker on openSUSE.

With openSUSE Leap 42.3 as base for the docker image everythings runs fine. With openSUSE Leap 15.0 its broken because every file and directory that will be created by PHP will not be owned by the apache process but at random user & group. Means after creating something the apache process has no access rights on it :frowning:

The stange thing: there is no SELinus or AppAmore active, apache has the right UID / GUID setup, and even a ‘whoami’ in PHP shows up as the right user & group from the apache process.

Maybe it is a simular problem ?

Best Regards
Thomy

  1. August 2018 14:46, “Andrew Erman” noreply@nextcloud.com schrieb:
9

I have the problem on ubuntu 16 with nginx. And I saw it with php 7.1 and 7.2.

10

for those about to test the beta2:

curl -s https://raw.githubusercontent.com/ReinerNippes/nextcloud13/devel/prepare_system.sh | /bin/bash
ansible-pull --clean --force --checkout devel -d ~/nextcloud14.beta2 -i 'nextcloud' -U https://github.com/ReinerNippes/nextcloud13.git nextcloud.yml \
-e "next_archive=https://download.nextcloud.com/server/prereleases/nextcloud-14.0.0beta2.tar.bz2" \
-e "fqdn=$(hostname -f)" \
-e "ssl_certificate_type=selfsigned" \
-e "talk_install=true" \
-e "install_collabora=false" \
-e "dhparam_numbits=1024"

it’s wise not to do this in a production environment. :wink:

using an aws ec2 (and other cloud server) you have to change $(hostname -f) to the external name. otherwise your browser won’t accept the certificate.

it runs best on ubuntu (16/18), on debian 9 and (without talk) on centos 7.

talk is just now installable.
collabora doesn’t run with selfsigned certificates. maybe someone can send me the trick to get it runing.

if you want to use letsencrypt:

curl -s https://raw.githubusercontent.com/ReinerNippes/nextcloud13/devel/prepare_system.sh | /bin/bash
ansible-pull --clean --force --checkout devel -d ~/nextcloud14.beta2 -i 'nextcloud' -U https://github.com/ReinerNippes/nextcloud13.git nextcloud.yml \
-e "next_archive=https://download.nextcloud.com/server/prereleases/nextcloud-14.0.0beta2.tar.bz2" \
-e "ssl_certificate_type='letsencrypt'"
-e "fqdn=<nc.example.org - replace>"
-e "cert_email = <nc@example.org - replace>"
-e "talk_install=true" \
-e "install_collabora=true" \
-e "dhparam_numbits=1024"
11

Hi guys,

several users have their install block on the table length issue - is there a way to unbreak it and continue/recover the install?

Just for those curious: this issue occurs when a table index name is too long. We added a check for this for Nextcloud 14 to improve database compatibility. Some databases don’t deal well with names that are too long!

12

Ok @nickvergessen had the solution:
delete or comment outthis line: https://github.com/nextcloud/server/blob/v14.0.0beta2/lib/private/DB/MigrationService.php#L460

The upgrade can be restarted, but you can’t use the auto-updater because it always redownloads and overwrites the change :stuck_out_tongue:
So you need to do the update manually.