Update Nextcloud to 27.1.5

[stondino00]

Fixes #2630

[Pilzinsel64]

The tests do fail at the same point then the daily tests. But only for NC27+. Not sure yet what the cause is. But this line sounds interesting:

In Setup.php line 568:
                                                                               
  /snap/nextcloud/x1/htdocs does not have enough space for writing the htacce  
  ss file! Not writing it back!      

Or it fails at this point:

maintenance:install [--database DATABASE] [--database-name DATABASE-NAME] [--database-host DATABASE-HOST] [--database-port DATABASE-PORT] [--database-user DATABASE-USER] [--database-pass [DATABASE-PASS]] [--database-table-space [DATABASE-TABLE-SPACE]] [--admin-user ADMIN-USER] [--admin-pass ADMIN-PASS] [--admin-email [ADMIN-EMAIL]] [--data-dir DATA-DIR]


An error occurred in a `before(:suite)` hook.
Failure/Error: DEFAULT_FAILURE_NOTIFIER = lambda { |failure, _opts| raise failure }

  expected: 0
       got: 256

But this topic is out of my scope/know-how.

[stondino00]

We may need @kyrofa to help on this one. Not sure either.

[kyrofa]

I suspect that the manual installation in these newer versions require write-access to the htdocs directory because it wants to write an htaccess file. The current version doesn't do that. We need to look into that, see if there's a way to opt out of that behavior. Sorry guys, that has been failing in the dailies for a while: those are intended to catch this kind of breakage flowing downstream before they bite us, but I haven't had time to investigate them, so now it's biting us. I suppose I'm the only one that receives notifications of those failures, huh. I should figure out a way to change that.

[Pilzinsel64]

I opened an issue on the server repo. Let's see if we get a feedback for this. nextcloud/server#42286

To be honest, I didn't find the commit that added that behavior. That are just too many commits there. :/

[Pilzinsel64]

I suppose I'm the only one that received notifications of those failures, huh. I should figure out a way to change that.

I did see it yesterday when I was looking for why the 28/edge builds aren't up. But I didn't bother as the tests for the PR I updated yesterday successed. I assumed that's just a temporary issue, sorry. :/

[stondino00]

Wonder if we are hitting this commit. This is new in 28 and 27.1.5

[stable27] do not write htaccess file if disk space is too low by @backportbot-nextcloud in nextcloud/server#41608

[Pilzinsel64]

@kyrofa Can you say anything regarding the available disk space for the tests on GitHub? Each nc release has a slightly increased size, maybe we just reached the disk limit?

The commit @stondino00 linked might be new but only throws if the free disk space is lower then 10kb (or MB? not sure).

[kyrofa]

Actually no, I think @stondino00 is spot on: the snap is a squashfs image, it's literally by definition 100% full. The available disk space on the host doesn't really matter here.

[pachulo]

Hey folks, do you think it makes sense to push 27.1.4 to the stable channel in the meantime?

[kyrofa]

Yes, I just haven't verified it yet, have you?

[pachulo]

I've just tested a fresh install and everything seems fine.

[Pilzinsel64]

An upgrade also works fine, tested on my prod.

[scubamuc]

fresh install tests completed and documented here

running tests again for confirmation...

---

logs look fine

[pachulo]

OK, in here they say that the latest updates solve some security issues, and they also now disclosed 2 moderate security issues that were solved in 27.1.4:

1. GHSA-3f8p-6qww-2prr
2. GHSA-5j2p-q736-hw98

So, how should we proceed @kyrofa ?

[kyrofa]

27.1.4 is in stable, now. Regarding 27.1.5, our only real option is to wait for a fix. We can't ship stuff that doesn't pass CI, and the manual-install feature is one that I know is used by people.

[Pilzinsel64]

Closing as we need to wait for 27.1.6 at least.