Zentyal 3.3 - A Magnificent Fail

[soorploom]

Z3.2, well, that was a disaster. A huge number of patches and it still couldn't serve files.

Where were the members of the Zentyal hierarchy? Nowhere to be seen on this forum with the exception of what would appear to be the usual expert suspects with their excuses and fawning admiration of the latest release, which, in the case of z3.2, is a complete brat.

Seen on this forum were comments to the fore that it was great at this, not so great as that, really bloody awful at some but overall, if you read the Zentyal hype on the main website, a real alternative to a certain other server package, as in "Easy IT for Small Business A native Microsoft-compatible, all-in-one IT backoffice".

It's not. Simple as that.

It's broken, faulted, incomplete. A server application which cannot serve files is no server at all. It has even been admitted by a Zentyal fanboy that he/she (mmm, maybe fangirl, then?) uses yet another box, another solution, to serve files. No 'all-in-one' bit there then.

After three months of struggling with the blood sucking z3.2, personally, I have, on one system, regressed to core 3.0.26, for the time being with absolutely no upgrades. File sharing is no problem, no hassle. It just works. At least, for me. My other system has now been changed to something else. That took a couple of hours but works as expected and required.

I have left a small system grinding away on z3.2, just for the hell of it. Definitely an S&M server. I thought, yup, give it a go so I followed the invitation to upgrade to z3.3. Much like others, apparently, it screwed. It even told me that a package wasn't fully configured, one that I had never installed, suricata if you want, and spewed out loads of dpkg errors.

Zentyal, no, nope, not going there again. For the best part of forty years in IT or what used to be termed as data processing, networks and the like, this z3.2 and its feral z3.3 offspring is a dog. Heard too many promises of super upgrades and promised bug fixes heading towards the unattainable nirvanah.

According to Zentyal, the Community version and the Corporate version are the same or if you watch the recent presentation, they're not or rather, won't be. Not exactly a great starting point to demonstrate the abilities of an unknown package to prospective paying customers as is exactly what I was doing. If you haven't seen the video and have absolutely nothing worthwhile to do as in absolutely nothing, try this: https://www.youtube.com/watch?v=gDQqkqkOJKo

It's dreadful.

I don't know how many paying customers Zentyal has but judging by the comparitively low activity on this community forum, those commercial clients must have no problems. At all.

[jbahillo]

Hello:

First of all I'm sorry that you experience with 3.2 has been so bad. there are some issues that should be fixed on it , for sure, but in general terms, the service works. I cannot enter in details as I cannot know what were your particular symptoms. Have you tried opening a bug report in trac?

Moreover I would like you to remind that at some times, it is not just Zentyal fault, but you might be mixing some issue from the software zentyal uses to provide a service and some zentyal issue.

As an example, let me show one you have suffered. When you mention:

I thought, yup, give it a go so I followed the invitation to upgrade to z3.3. Much like others, apparently, it screwed. It even told me that a package wasn't fully configured, one that I had never installed, suricata if you want, and spewed out loads of dpkg errors.

Well, you might be interested in knowing that this is a bug in suricata package (not on zentyal-ids but on the upstream suricata package) that has been detected by zentyal staff and notified to ubuntu community: https://bugs.launchpad.net/ubuntu/+source/suricata/+bug/1250439

if you still want to give a try to zentyal 3.3 you can try using the patch to workaround that particular issue

[Doward]

The fact is that Zentyal is marketed as a simple, easy to use system.  It is - WHEN it works properly.  Unfortunately, if something goes wrong you need *quite* a bit of Linux know-how in order to find, decode, and fix the issue.

Unfortunately, those level of SysAdmins are not the 'target market' for Zentyal.

IMHO, the Zentyal team needs to spend some significant time cleaning up and reenforcing the Error subsystem.  Give the users simple, easy to understand reasons for their problems.  Better error checking, responsiveness, and control is absolutely needed in this project.

Just one man's meager opinion.  I use it for the front end interface, but I can tell you that my Zentyal system is completely virtualized.  I make no 'upgrades' without first checking them on a VM clone.

[christian]

It's broken, faulted, incomplete. A server application which cannot serve files is no server at all. It has even been admitted by a Zentyal fanboy that he/she (mmm, maybe fangirl, then?) uses yet another box, another solution, to serve files. No 'all-in-one' bit there then.

The way you report it clearly shows that you have truncated reading or understanding. I know some Zentyal users running something else than Zentyal as file server. BTW, that's what I'm doing too. Why would you mean or state that reason behind such design choice is that Zentyal file server doesn't work ? 
Speaking about my own choice, as I'm still run 2.2 that is based on Samba 3, my Zentyal file server is quite stable  and reliable. It just doesn't fit my needs, reason why I run something else 
I can easily imagine that other Zentyal users having decided to run another file server have their own reasons too, like ZFS support or need for NFS...

[Kurtis]

Don't know if it has been mentioned before but if there was a was an easy way to rollback to previous updates if a update does cause the server to break then that would be awesome

[robb]

Don't know if it has been mentioned before but if there was a was an easy way to rollback to previous updates if a update does cause the server to break then that would be awesome

Actually there is:
On physical systems
Get your system on a decent Raid1 volume
Before you upgrade shut down the server, break Raid1 and startup the server again
Apply patches and updates and testdrive on a broken Raid1 for a day or 2.
If ok: sync
If not ok, down the server again and switch the drives. Rebuild the 'old' (pre-update) Zentyal version.

On Virtual systes:
Take snapshot
upgrade
When ok, be happy
When not ok: revert to snapshot.

Something tells me this is quite basic upgrade/patch procedure that should be standard practice in a professional ict environment.

[half_life]

One caveat to Robbs suggestion,  either method requires that you address what to do with email.  You may shutdown the mail system or plan how to transport the email changes back to your backup system. 

Roundcube stores email in /var/spool/mail
Zarafa stores email in a mysql database

Personally I just type in

Code: [Select]

sudo /etc/init.d/zentyal mail stop
to stop incoming email until I am satisfied that the upgrade went well

Another option is to just ignore it and risk losing email.

[jjans]

I gotta side with soorploom on this one....
Core 3.3 'may' be ok with new domain,  but I shudder to imagine what would happen if I actually migrated over from my SBS 2011 production environment. (is this not the point of Zentyal??)

All I did was attempt a simple upgrade from Zentyal 3.2 (running zarafa) on a test KVM environment and the installation of openchange simply broke everything:
* did not enable openchange account for existing users
* did not enable openchange account for new users
* mapiprofile could not even connect to itself to create a profile ...
* could not revert back to Zarafa.
* could not reinstall Openchange and reprovision the First Organization (or a new 2nd one...)
* could not get webmail to work...

This was a simple upgrade of an existing core 3.2 environment that was pretty much 'out of the box' with some users and a couple xp clients (also VM's) - nothing fancy, just as per the 3.2 tutorials / documentation.

Only upgrade scenario that worked for me was to shutdown the original domain, and create a new one out of the box with 3.3 (and then create new users matching the original...). Then importing previously exported PST files into Outlook seemed to work OK after rejoining the client machines...

Don't get me wrong, I still think this has potential, but documentation of upgrading from Groupware is lacking, and not even mentioned here -> http://trac.zentyal.org/wiki/Document/Announcement/3.3

[christian]

All I did was attempt a simple upgrade from Zentyal 3.2 (running zarafa) on a test KVM environment and the installation of openchange simply broke everything:

On the other hand, OpenChange is still very new and integrated to Zentyal only few time ago.
Plus 3.3 is not exactly a stable platform isn't it? 

[jjans]

All I did was attempt a simple upgrade from Zentyal 3.2 (running zarafa) on a test KVM environment and the installation of openchange simply broke everything:

On the other hand, OpenChange is still very new and integrated to Zentyal only few time ago.
Plus 3.3 is not exactly a stable platform isn't it? 

Yup - right you are. I will attempt migration from a test SBS 2011 in  a virtual environment and see how that goes... Think I got my hopes up a little too soon here...

[Lonniebiz]

From watching the video, I got the impression that releases 1.0, 2.0, 3.0, and 4.0 are the same for the community as for the paying customers. And "releases" in between (such as 3.1, 3.2, 3.3) are community "releases" that I myself would call "beta releases" rather than just "releases".

So, I've gathered that these in-between "releases" are really just "betas" that Zentyal wants the community to report bugs on so that 4.0 will be solid for paying customers.

I think Zentyal could prevent a lot of frustration, from new community users, if they would use the term "beta" for all "releases" that have a number other than zero after the decimal.

I started using Zentyal when 3.0 came out, and it lived up to everything promised. When other "releases" followed, I expected them to be just as solid as my first experience with 3.0, but from watching that video I see that I just happened to start using Zentyal on a .0 release. 3.0 and 4.0 are expected to be more stable than anything in-between. That video was helpful for molding my own expectations of Zentyal, but they should make the risks more clear regarding these in-between-major-releases.

[robb]

I see a much bigger problem arriving:
Until now the community version was very well able to become a solid solution in a SME ict environment. With the introduction of the new release schedule, more or less forcing community version to be upgraded every 6 months, I fear a lot of users will back off from either using Zentyal, or in a less worse scenario, refrain from upgrades and stick with LTS untill a new LTS version comes out.

The latter will be an opt-out for (beta)testing and instead of gaining a lot more test results, Zentyal will loose a lot of test results.

I absolutely can't see what good a _server_  solution brings on a 9month supported system. And why spend time on coding for a 9month supported version while the commercial version needs the LTS version code anyway. Why not stick with the LTS version for the community version?

I just don't see the benefit... (technically spoken of course)

[J. A. Calvo]

robb, the advantage of using the last Ubuntu release (wether it's LTS or not) for community, it's to anticipate the migration from one LTS to another, doing it in a much more progressive way. From one LTS to another there are 2 years of difference, and that means A LOT of changes in the base system of Zentyal, this new release policy is much less painful, because changes from one version to another are much less. And the final user also benefits, not only the developers, because there is much more time to test those changes when the new 4.0 based on the new LTS arrives. For me it's a win-win situation. Of course if you are using Zentyal in a critical environment and you are more confident with the version based on LTS (3.2 in this case), you are free to stick with it

[robb]

Josh, thank you for replying on this topic. I think it is very important to discuss this and see both sides clearly: Commercial decisions vs community needs and expectations.
Unfortunately the community has been left aside the last year due to IMO commercial priorities. These priorities are absolutely understandable. A project without marketing and devs is no project and marketeers and devs cost money, a lot of money. However, I also do believe that an opensource project can not survive without a solid community. This community also has its needs and those have been neglected last year

on topic: Historically the upgrade options from one Zentyal version to another are rather painful (to say the least). What guarantees will the most important version of Zentyal (as Julien stated in his Summit presentation) get when it comes to upgrading from one version to another?
You do realize that a 3 month upgrade path will degrade that most important version (the community version) to just a testing version and no sysadmin in his right mind will put that into production. This also means that absolutely no production testing will be done with the community version of Zentyal.
And please don't make the mistake that current community servers in production environments will be transferred to subscribed versions because of this change. It will not happen.
Sure, some home servers will stick with community edition, but every community server that is currently in production will cease upgrades until the next LTS version arrives. At least, that is my best guess and how I would schedule the upgrades.
I am absolutely not convinced the choice of a 3month upgradepath is the right one for a _server_ solution.

[christian]

I tend to share Robb's view here:
if there is no strong improvement in the upgrade process (and there is no reason to believe this could improve as Zentyal dev team will even have less time to test some many releases per year), very few Zentyal users will dare to upgrade their running platform thus benefit of "Zentyal releases tested by Zentyal community" will vanish.

On the other hand, (some) community users are pushing to get always more and more features, especially the very last one found on the net  and this has, for them, priority over platform stability