definitely. especially the handling of desktop client 3.4.0 I’m really disappointed…
I don’t do either… but looking at this thread and few others 12 shows similar picture - information flow from Nextcloud GmbH to the community is very limited - I would expect the vendor to take part in such important generic discussions…
I’m sorry if my word are not clear: my idea was to show the huge motivation of the log4j developer who works all throughout the day for free in opposite to Nextcloud which seems financially successful and still don’t mind completely fix the catastrophic issue discussed in 1 or at least don’t share it with the community. I’m wondering if some of the paying customers where hit by the issue and how they solved it? I can’t imagine they performed a server restore…
I guess they don’t use always the latest versions and features. But I don’t know anyone who has an enterprise contract, so i’ts only a guess.
Maybe it’s not that easy for them to solve… And I also get the impresson that many of the issues with the desktop client mainly appear on Windows together with Virtual Drive, which relies on features provided by Microsoft. At least there were fewer issues before they intoduced that feature. Personally, I never had issues like that with any version of the desktop client, neither on Linux nor on Windows. But I use mainly Linux and never used Virtual Drive, so I don’t know… But maybe they need better Windows developers, perhaps conceptual mistakes were made or maybe the documentation and support from Microsoft is poor and therefore they struggle to integrate it reliably. Most likely it is a combination of more than one factor. Anyway I can understand that you are frustrated. I just wanted to put the whole thing in perspective. And as you said yourself: You can’t really expect someone to work 24/7 like the Log4j dev does.
windows 11 here on 1 x desktop and 2 x laptops, heavily reliant on windows sync with the 3.4.2 client, zero issues before or after the 23.0.1 update, I do make use of virtual files too on one of the laptops to save space, still working fine so far after 24 hours
Maybe it leads to the point that the developers start testing their new releases before letting them out to the public?
And when I say “testing” I mean REAL testing; on a new machine; on a machine with an older version of NextCloud and so on - just professional testing, and not of he kind: “When it compiles it’s finished.”
OK, that may lead to even more delay. But for my part, I am more interested in working software than in as many updates as possible.
I am using NextCloud for years now and I can only remember one single upgrade that didn’t end up in a complete new installation and configuration.
Right now my (Linux) client tells me that V23.0.1 is out, but when I go to my server it tells me, that V23.0.0 is the current version. If not even these basic things work, how can one recommend NextCloud to anyone who just need a working tool for their synchronization purposes.
As of now, Nextcloud is a nice toy for nerds that have plenty of time to play around.
they did. There even is always at least one official RC to be tested before release. The question is: why don’t enough user take part in it? or even more specific: where have YOU been when it came up to betatesting or rc-testing?
NC isn’t a software just for rookies… you need to know what you do, when and where. I wouldn’t recommend it as a install & forget software. it needs to be maintained permanently. And no, you don’t need to be a “nerd”… If you can’t do it yourself let others do the stuff for you… either by chosing the right install-procedure or by buying support (like in: there are hosters offering preinstalled instances)
no every kind of OS was affected Windows with/without VFS and Linux (unsure about Macs). Jospoortvliet confirmed they don’t use Windows lot but the issue hit even their employees.
But the main point is not this issue happened - better testing may or may not have prevented this, the biggest issue was almost not existent problem management! the only official guy participating in this forum and taking all the bad mood was Jospoortvliet - a marketing guy without deep technical understanding - and later when the size of the problem was recognized there was no real push to fix it… only the client engineer published few basic attempts to reset invalid date to current time. at least in my case all broken files still existed as older versions - it can’t be such hard to create a script/describe manual procedure to mass recover existing file versions based on specific criteria…
but back to the current topic: I really hope this weeks without updates result from learning and better internal testing before updates become public.
yes I know this script and worked through the script but I found no attempt to recover original mtime… it only iterates through DB and files… additional flaw is it doesn’t touch files in case DB mtime deffers from file system mtime…
mtime might be not the most important data but depending on the files e.g. your family pictures bad mtime could be really annoying
in some maybe not all situation there is a way to recover the mtime from versions file (or it’s DB record).
I didn’t find time to build reliable recovery procedure from this information as I’m not an expert in mysql but it should work as following:
find files with invalid mtime
find the most recent version (if many) of the file which has valid mtime
perform one of the possible recovery options:
safe: touch the file with valid mtime (and remove version in 2nd step)
storage efficient: move version file to the original location and overwrite bad file
perform occ files:scan all
additional sugar:
compare the version file you use for recovery with the file in the system (diff/checksum) - AFAIK the contents are unchanged, so file contents and versions should be same
remove newer versions having invalid mtime (there are github reports when upload with invalid mtime happened multiple times)
I didn’t investigate further - NC devs might know from the top of the head - i think oc_filecache table is not separated by user so if collect the information from DB you don’t need to iterate over each user to identify and recover the files, it should work in one go…
A customer is someone who pays for something, there are not two tiers of customers, there are users, and there are users who are also the customers of Nextcloud Gmbh.
I have no connection to the company, I have my own small company where we use Nextcloud, for free (for now). As my way of appreciating the project I have submitted and had accepted the occasional pull request, and attempted to supply high quality bug reports on github.
May I ask what you have done to support the project so far that gives you the moral right to make such demands?
I can tell you now, without the paying tier of users, Nextcloud development will slow considerably, so you better hope they keep making those paying customers happy.
I’m using it from the beginning and had only once a problem with an app that was not ready for the new version yet. So experiences can be very different, some setups and some functions and apps are much more problematic than others.
I don’t think that they take it easy like that. It is hard to test everything for everybody.
However, problems shouldn’t happen and to blame everytime that it is free software and you don’t pay for it doesn’t count either. With that, we need to test Nextcloud ourselves. But to run Nextcloud we need Apache, oh that is free software too, so we have to test this as well. MariaDB, …, Linux? At some point you want to just use the software, if you have to do all the debugging and fixing, you never finish.
Question is, how can we improve that. On one side the community, that can do certain testing, give feedback etc., and also the company.
Regarding new features, perhaps keep them longer or more clearly as a alpha/beta feature that has to be actively enabled. It’s no problem if the community helps to test, but they should be aware of it and perhaps not use productive setups, and it would be helpful what has been tested and what still needs to be tested. Like for new features, if the developers didn’t have any Mac systems, some Nextcloud users with many Mac-systems in use could be happy to do certain tests within their systems.
sounds like you didn’t spend a second to check my forum profile before you offended me without a reason.
Next time when you decide to attack somebody please try to understand the situation completely. if it happens I bump your car - I will tell you it’s worth no reaction as you have no contract with me - deal?
everybody has moral right to expect a damage to be recovered. the case I’m talking about is pretty easy - software provided by Nextcloud caused huge issues for many users - and the reaction was far not ideal.
ACK. I started using NC 18 months ago. I run 3 major upgrades with small issues (on two instances - I use one for testing). I have to admit I spend maybe 3-4 months to find a good setup for me, tested backup and restore, decided which apps to use and which not. Even two cases when client removed files or issues bad mtime was only annoying and caused some work - but there was no data loss. I agree same scenario with lot of users, business need might result in lot of stress - but even it’s possible to build a process to recover from such scenario (if you rely on software you may be you should start building this process from the beginning)
Fully agree… unfortunately at the moment it looks like there is no good testing at dev side… maybe they are overwhelmed with customer support, maybe the development is too fast… fact is - if a dev misses test case happening once in 3 years - no blame, when a dev misses a situation happening once in 3 minutes - bad mood…
longer testing periods might help… but majority of the users don’t wan to test, they want stable software - longer alpha/beta testing could just postpone uncovering the issues until there is no professional guidance like "users with Mac OS 123.654 please configure the client with this feature and run following test cases… "… Leaving “testing” to untrained users and hobby admins leads to exactly the situation we have: “tested” software is released to public stable channel and cause issues and data loss and in turn bad image for the product and the company
@tflidd and @wwe - so how long would you think the testingphase should take?
afaik the releaseschedule for every new major version sees 6 weeks of testing… including 3 beta test stages and 2 RC test stages until the final release (with a feature freeze on beta 1). so this IS quite a long time for testing regarding that we usually end up with 3 new major versions each year (some circumstance I personally was questioning since long)
PLUS we absolutely don’t know how hard the pandemic hit NC - devs. Like many other, bigger, companies are fighting this problem as well, offering less service, shrinking down their offers, shifting projects and all of that. And everyone here is freaking out about a postponed release?
@wwe I saw the discussion about the one nagging bug. And I do appreciate your qualified input there very much.
the bug looked solvable. But I’m no coding guy… Though somehow it really wasnt’t solved until the other day (didn’t look in there since then). I dunno why it’s still unsolved. Maybe they don’t have enough devs for it? I dunno. I can remember that especially this “virtual” seemed to be a lot harder to code than everyone expected and thus ending up in a way longer development for that feature. Maybe fixing the bug seems to be harder than everyone thought? But apart from that… this thread started asking why NC23.0.1 was so heavily delayed and now we’re discussing problems of some totally different app?
Do you remember when I asked here on this thread where the discussion would lead? Now we know. And to be honest: I don’t know any relevant piece of information more than from the beginning on. But well… this discussion left some of us in a bad mood, with bad feelings, of course. Was that the goal?
I can only repeat: I am sure that all devs do whatever they possibly can to offer a good product to us, regardless of “us” being user or customer.
Unlike Nextcloud I never encountered any problems with Apache, mysql or MariaDB.
I also think that when there is a missing database field after the upgrade of a vanilla NextCloud that doesn’t look like a problem with the environment. I’m using PostgreSQL for my single user instance. Because of the update problem, I don’t dare offer nextcloud as a service to my friends.
I would like to see fewer new features and more consolidation of the existing, knowing that this is far less interesting for developers. And this doesn’t just apply to NextCloud. It’s a worldwide featureitis virus that makes modern software more and more a pain in the ass.
But at least, I never had any data loss. Just got stuck on updates causing me to save all files locally, saving my mobile data (contacts and dates with MyPhoneExplorer), deleting the whole stuff and setting it up completely new.
That’s not something I would like to do when I had some or even many clients on my cloud. For me alone it’s not a not catastrophe, but pretty annoying.
