Something goes wrong in this forum when discussions on the same topic repeat themselves

Something goes wrong in this forum when discussions about serious NC upgrade issues continue for months when the solution has already been given.
Surely there must be a way to intelligently highlight an important problem that many users are facing in combination with the best solution. A kind of up-to-date upgrade/configuration overview. Especially when it concerns problems that have been reported to Github.
Here is an example, which has been going on for months! The now famous PHP-apcu module (memcache) settings and cron jobs failure.

An onter example: the problems with desktop clients v3.2.1 and 3.2.2 now solves with the fix in v3.3.3.
Many thousands of users may have faced these 2 problems and spent days looking for help. In my situation it was solved quickly, but this was pure luck.

Well such configuration changes like "apc.enable_cli="1 should maybe be highlighted in the documentation. But in this specific case I think that this parameter was already present in the documentation, but only on day X with a certain update really led to the fact that the cronjobs no longer functioned.

Anyways… The main challenge with your suggestion is, that you can install Nextcloud in about 1000 ways. With or without memcache, with APCu or another memcache. With Apache, With NGINX, with other webservers, which can be configured differently, with different PHP versions, on different Linux distros with different versions of the required components etc… How do decide what to highlight? Mabe they could maintain a changelog for the most common configurations, but even that won’t catch all problems because as I already said. There is not the one right way to install Nextcloud. And many people copy & paste from random tutorials without knowing what they are doing.

As an inexperienced (home) user, you have basically two options:

  1. Use an appliance like Nexctcloud Pi or the Snap Package

  2. Use the manual installation methods described in the documentation (example installation on Ubuntu or CentOS) and then check the documentation for changes, before you upgrade anything on your instance.

To maintain a manual installation, you have to aquire at least basic knowlege about the components involved, so that you can help yourself to some extent. Most things are documented on, but still require some basic knowledge about Linux, web servers, PHP etc… In addition to that, you should keep your own documentation and document every change you made on your instance.

If you use Nextcloud productively in a company, there is really no excuse if a change like this causeses downtime. In commercial environments, you absolutely have to test upgrades before rolling them out to production. Anything else is unprofessional and as an admin, you can’t blame anyone but yourself, if you didn’t test.

And one more thing…

You don’t always have to upgrade to the latest Netxloud major version, PHP version, MariaDB version, OS version etc… on day 1. The last two or three Nextcloud major versions still receive security fixes. Ubuntu 18.04 is still supported for two years, MariaDB versions are supported for 5 years. No need to imdiately upgrade, just because some commponent has a new version number.

My production instance is still running PHP 7.4 on Ubuntu 18.04 with MariaDB 10.3. I will probably upgrade to PHP 8 and MariaDB 10.6 with Nextcloud 23 , meaning all these PHP 8.0 and MariaDB issues, that are discussed all over the forums, will never have affected my instance. By the time I will migrate, everything is well documented and tested. And still, I will run at least the database migration on a test instance first.

A few additions that are a more direct answer on the actual topic of your post, while I still think that is the appropriate place for most things you mentioned in your post.

First of all, it’s not the forum’s fault if people don’t use the search function before posting :wink: On the specific topic you linked to, several new posts came up on the same day, while the original post still was in the first 20 or so posts. Even a sticky would not have prevented lazy people from opening a new thread then :wink:

Your idea is not bad. But that would have to be actively curated and updated. Who desides what topics to pin? How long should they stay pinned? How many topics should be pinned at the same time? Too many at once will defeat the purpose. But even if the mods would do this extra work, you could not catch everything…

First, I think it would be helpful to have a category that deals with upgrade and ‘business continuity’. After each new NC stable release, this could contain a link to change management, against which recommended configuration is ‘stable’ defined?

Secondly, it would also be frankly stated that when users deviate from recommended configurations, no priority is given to forum support in the above proposed category, i.e. you are then on your own. In my opinion, the ambition to make NC an Enterprise robust platform also includes taking small and medium-sized companies seriously and that stable releases are stable.

Thirdly, if unexpected conflicts arise, in the proposed category the problem description + workaround could be given a pin until an interim fix is made available with a reference.

The search function has the characteristic of not always being helpful when the basic problem is not yet clear. Which keywords should be used? Once this has been sorted out, the keywords are clear. This is a chicken and the egg situation. I wouldn’t call forum visitors lazy, after all, a good number of members with full agendas come here to actually solve a problem because something doesn’t work and their business, relationships and friends are suffering.

The idea is good. But I still think it would be difficult because, as I said, there is not the one correct configuration. And in this specific case it was not a bug, but only a change. But maybe we could maintain a list with breaking changes with the most common configuartions and dependenceies, together with possible solutions or links to the documentation, as they appear…
The question is, who would do this work? Are you volunteering? :wink:

If you are an Enterprise you can buy support from them…

You can minimize the risk of breakage with a test environment and by upgrading only after you have tested the new version. This is common best practice, also with commercial software. If you want someone to do this job for you, you won’t get around paying someone to do it. While Enterprises usually have an IT department for this or pay the vendor directley, smaller ones usually pay an external IT service provider. And home users can use one of the appliances that I mentioned.

That sounds like an issue tracker and not like a forum. :wink: Again my question: Are you volunteering?

But I agree that maybe the bigest issues, that affect many users and arrise very often could be pined for some time by the mods or put in a “known issues” category ore something like this. However, one must always remember that all forum participants, including the mods, do this in their free time on a voluntary basis. Nobody gets paid for contributing to the forums.

It’s a shame you don’t get paid and I’m afraid this is also a statement from Nextcloud GmbH that companies with less than 100 employees are excluded from professional support. The entry number for an Enterprise support subscription excludes small and medium-sized companies. Say like Nexlcloud itself with 35 employees.

Then it would no longer be voluntary, but a job. I don’t know if I would still find it interesting to post here. The expectations of the people would be different and it would no longer be a real community forum.

I don’t know the exact reasons why they decided not to support small businesses directley. But I think part of the reason is that they want to focus on developing Nextcloud. And that they are a relatively small core team. And I personally don’t think that’s so bad. I’d prefer them to put as much resources as possible into the development of the product instead of dealing with end-user or small business support. In the end, everyone benefits from this.

That particular case was indeed a bug at first and then a suboptimal workaround for an issue with should not exist. So there was no preparation, no info about a breaking change or prominent changelog entry. Please help me fix the underlying issue: Fall back to NullCache on CLI calls if configured cache is not available · Issue #27608 · nextcloud/server · GitHub

“Stable” setups are well defined in the docs (which included the ACPu hint as well btw) and the official appliances. For any important changes, read the changelog instead of hoping for pinned forum topics. In this particular case it was simply not known.

This is a community forum where volunteers help others, so there is no priority that could change. Paid support is a different thing of course.

Stable releases usually are stable and breaking changes usually are documented in the changelog. But noting is 100% which is why not upgrading at day 1 and other mentioned measures are reasonable when you rely on the uptime of that Nextcloud instance.

I like everything said in here…

However I need to disappoint everyone looking for “pinned” threads… It won’t help in case of certain nagging questions…
Meaning: you can pin something at the top and though there will ne new threads asking for exactly what you just pinned. :man_shrugging:

For me I just lost hope that some users will just once fill out underlaying templates or use searchfunctions or would get away from the idea that they will get immediate and individual “support” on here…


It is a special experience for me as a new participant in this forum that the topic of the discussion I started, evokes so many reactions that ‘nothing can be changed’. Apparently, either the problem is not clearly perceived or the forum maintainer or moderators find the subject difficult. The easiest thing to do then is to blame the users, the search engine is not used properly or everyone wants something different with different expectations.
I invite everyone, especially from Nextcloud GmbH to formulate what they expect from this forum and what purpose it serves. It has probably never been thought through properly. The contrast made between Enterprises and home users is also very painful. It is forgotten here that the most effective NC promoters, which Jos Poortvliet needs so much, are in the small and medium-sized companies. People from these companies enter all large Enterprises. Whether they have an on-premises installation or work through a solution provider, this group is totally neglected.


you are “new” here as you said yourself. So how do you know what mantainers or moderators has tried already? Where do you know from what has been already talked about up and down?

The bitter reality is like it is: of course there are users using the searchfunction. But there a good few that just give a :poop: - even if you remind them about using it.

They did that already: It’s a community-driven forum living by volunteers trying to help. Every bigger Open source-project has that. All NC GmbH does is giving out the technical basis to run the forum. And at times some devs will come here to look what’s going on and give a helping hand.

In the contrary… it’s well though through and NC wasn’t the first company doing so (they only live from company-users paying for support!) nor will they be the last one. It’s common!

There’s support for them as well… they just need to look for it. It’s only provided by 3rd party companies. (and at times THEIR employees do come here asking for support as well… )

You (and others) might have forgotten that NC itself is Open Source. Which means: everyone can have a word - if they like. Everyone can code - if they dare to and everyone is asked to get their knowledge in as good as they can.

NC GmbH isn’t the owner of NC. It’s a company offering support for this product and hire devs for the NC-project to do some coding.

home users need to rely upon the help they are getting here on the forums. That’s how a community project works.

NB: Btw … we even had this kind of discussion several times before here on the forum… :man_shrugging: , funny, eh?


Just to answer the initial post, we had this in the past when it still was ownCloud (e.g. Upgrade to oC 8.0.x (known issues) - Server - ownCloud Central). However, they improved the update procedure quite a bit and it got much more reliable that such huge general topics weren’t necessary anymore.

This APCU issue was a major problem that would justify such a sticky topic again. Normally we started with known issues and moderator did add stuff over time. We need to keep an eye on that and might do that next time.

Now, it’s probably a bit late for the current version.

1 Like

Well, it’s easy to blame others. Look at the number of topics here with many basic information missing (page is just a white screen, internal server error, …). You never know what they did behind, if they just post this, did they do some research but were overwhelmed by different topics, … and for moderators, it’s always the same game, did you check the logs, etc.

In the howtos and FAQs, there are some nice write ups of common problems, we can easily link. But doing these, takes quite some time. Perhaps we should invest more time in that than answering recurring questions and present these tutorials/FAQs more prominently. The modern forum design makes older topics disappear quickly even if they are still of interest.

Everybody should keep in mind, that all others help out in their free time. So if some expect to get some support where people here in the forum invest more time in resolving their problem than they do, there is a problem.