[Proposal] New way to handle upstream updates

@flatspin,
I understand that we will never be 100% perfect but that should not be used as an excuse to hold us back from trying.

2 Likes

maybe you misunderstood me… my time, your time, everyone’s time is valuable… if we have to invest it on the product, we’d rellay invest in something that is new, unique, not just rewriting something someone else already did.

I’d prefer to spend 2 hours a week to test and debug something new, instead of copy and paste Centos’ documentation to document something useless (already existant)

hope my POV is clearer now

1 Like

That’s right, but to be also a little philosophic:

If you reach a certain level, the effort to getting better grows exponentially and at some level it doesn’t make sense to invest more, because the investment doesn’t stand in any relation to the improvement.
The question now is: Are we allready at that level? :wink:

3 Likes

@Stefano_Zamboni,
This is a large and diverse community (I use the term “diverse” in a broader meaning and not within the current limited politically correct / corporate speak definition).

Sure, you may not have time or the inclination to examine the wider issues related to this subject, but there are others that would like to see a more integrated approach to these related issues.

Remember, no man is an island, we are all part of a community with similar / overlapping aims and objectives.

4 Likes

probablement à l’insu de mon plein gré (french joke)

More seriously, what this module needs more ?

2 Likes

Today @giacomo and I were thinking about this solution and its feasibility. How to implement it?

We found two alternatives.

1. Software center banner

As upstream problems occurred with minor releases (7.3->7.4), we could display a banner in Software center page that explains what will happen if the system is updated. In other words, instead of the usual yellow “Updates available” banner, we could display a scaring red one: “System upgrade”.

This solution is an enhancement of the current policy. It does not implement the fast/slow policy but still leaves the responsibility of updating the system to the sysadmin.

It provides a better awareness of what’s going to happen (hopefully).

2. “Slow” policy mirror infrastructure

The fast/slow policy requires a set of additional mirrors to serve the “slow” updates. Whilst “fast” would rely on the upstream mirror infrastructure (as we’re doing by now), the “slow” must be implemented with our own servers.

This solution seems quite expensive because upstream mirrors are on a different order of size compared to our tiny NethServer repositories. Disk requirements (yum repolist -v) are the following

At minimum:

base, updates / 26 GB

But also sclo and extras should be mirrored for completeness:

base, updates, sclo, extras / 40 GB

Moreover, the less mirrors, the more bandwidth/load is required!

What do you think? Who’s available to host a such mirror? Isn’t the banner enough?

7 Likes

Of course not :slight_smile:

But please guys don’t go off-topic, we should focus on what we can do NOW with our repositories in order to fix the problem turned out at the conference, as @giacomo correctly explained.

2 Likes

A post was split to a new topic: Talking about a clear “managerial structure”

I was pointed to Proxmox subscription plans and this thread came to mind. We could cover the slow updates repository costs with a paid subscriptions plan.
https://www.proxmox.com/en/proxmox-ve/pricing
Probably worth a new thread @alefattorini.

You make me think the two “alternatives” aren’t mutually exclusive.

We can implement both the banner and an alternative (slow/managed) updates channel!

2 Likes
------------------------------------------------------------------------------------------------------------------------
   74,4 GiB  /smeserver                                                                                                                   
   60,8 GiB  /archlinux
    3,1 GiB     /nethserver
  733,3 MiB   /stephdl

actually the 4 repositories hosted on my mirror

3 Likes

While we can keep going the discussion on a possible paid subscription plan, I would like to have some firm points to proceed with next NethServer release.

If anyone agree, I’d go with:

  1. Define a list of hot point to check before releasing minor release. Please add here: https://wiki.nethserver.org/doku.php?id=howto:qa_testing_hot_points (@quality_team, @iglqut)
  2. Add the banner inside the Software Center to inform users about upgrade between minor releases (as suggested by @davidep)
  3. Move @stephdl’s yum-cron package to the core, so anyone can enable automatic security updates (please bear in mind that a security update could also disable insecure feature causing incompatibilities usually with legacy clients)
  4. Better document update policies as request by @medworthy (@docs_team would you like to create a new wiki page or adding directly it to the manual?)
6 Likes

I think it belongs to the manual- splitting up to much makes things more complicated, than they already are.

2 Likes

I think this is the solution we allready have, a community version with fast updates and the enterprise with tested updates a little bit later.

2 Likes

What do you think about checking modul function like squid, groupware and nextcloud? I think these are moduls which are used very often.

1 Like

Here’s a preview

Any hint?

2 Likes

I agree, could you add it to the linked wiki page?

Excellent, this is a “go” for me :wink:

Of course. It’s added.

2 Likes

I think we are ready for releasing NS 7.4 final as all hot points have been tested.

Just documented the update policy as requested: please review! (/cc @docs_team @medworthy)

Last bit: would you like to have nethserver-yum-cron package inside the NethForge?
If yes, @stephdl has the last word and should move the GitHub repository under NethServer organization, then I will gladly publish the package in NethForge for 7.4.
But we have no rush, we can do it after the final release or leave everything as is: users will be free to join Stephane repo! (I like this one ;))

2 Likes

I fully agree. Let’s talk about paid plans and community servers first.

2 Likes