I used to wait for v7 for some time (years, so to say) for some thin features like BGP performance, and while I can live without, it was nice to know one day it’ll be there for us.

But as I cam observe today, it is a bit strange situation for us to consume:
- all new features are to be addede to v7 only (which is logical),
- all new hardware are to be used under v7 only (new chips etc)
but
- v7 still not even a beta, just RC (for years), while all of use are familiar how “stable” is beta releases.

So, I just can’t figure out, If I can rely on v7 for current deployment (today I want to see OpenVPN and ZeroTier as production ready), and understand when will it be released so we can put it into enterprise usage.

What’s your opinions, if I can trust v7 (of its current stability) today (that is, at least it won’t break the network and traffic), or I still have to wait (another 2-3 years?), and not to buy new routers that’s v7-only?

TL;DR no. stick with 6.x

It's a thing that's bugging me for the last couple of years now. Mikrotik software quality really deteriorated since ~6.40. Even had a mishap when using 'stable' version with normal upgrade path when SFP+ just failed on couple of RB4011 and really borked my vacation. Since then I'm really careful with every update. All automatic upgrades all long forgotten and I sticked to 'long term' releases for quite some time now. I upgraded about 50 boxes to 'stable' about a month ago. Today, a new 6.49 with a looot of new features was released and I'm just too damn scared to use it in production. I'll wait for couple of weeks to check some forum feedback if everything is nice and dandy and prolly upgrade afterwards.
I do have some CHRs for non-production stuff that are running 7.1beta4, mostly for wireguard as it really is fast, stable and not too chatty.

TL;DR no. stick with 6.x
I do have some CHRs for non-production stuff that are running 7.1beta4, mostly for wireguard as it really is fast, stable and not too chatty.

Thank you! The same feeling on my side. I actually need few CHRs to connect several rented servers in different data centers and 5-6 MT router boxes (pure routers, no fancy wifi all-in-ones, so less code there to see bugs) in offices, and to use v7 and ZeroTier seems to be nice way, but I so scared to upgrade longterms to (oh, my!) rc!

The bad thing, there are only a few alternatives that I can see (and WRT on MT hardware is among them), so v7 is a option.

Nobody else can do testing for your specific setups, but only you! I run v7.1 in several setups without any problems and in some other setups it wasn't that successful, so there I don't use it. Easy as that.

It failed me today. So, no.
Maybe tomorrow.

v7 is in RC stage, so it says right in the name it's not meant for production (or enterprise).

In other cases, it depends on amount of time you are willing to spend to deal with debugging alpha and beta bugs. In my case, even at home that amount of time is zero, so I use LTS ("long-term stable") branch and even have a bash script that performs the upgrade only if LTS version is older than 30 days.

v7 is in RC stage, so it says right in the name it's not meant for production (or enterprise).

In other cases, it depends on amount of time you are willing to deal with debugging alpha and beta bugs. In my case, even at home that amount of time is zero, so I use LTS ("long-term stable") branch and even have a bash script that performs the upgrade only if LTS version is older than 30 days.

I used to run 'Long-Term' releases releases as well, since this is something we can rely on in terms of stability. Sadly I saw situations when even 'stable' branch releases revealed different small problems - then, right as you've written, someone (me) has 'to spend time catching bugs'. In enterprise, even small one, if time you use to drive to remote branch office is not zero then dealing with beta (so called "stable") branch, not mentioned "testing" (read "semi-alpha") or "release candidate" (read "we don't have time to test it internally") is crazy.

The problem is, MT seems to slow down its software development and put most attention to 7.x branch, so 6.x seems won't get any new features (like ZeroTier), but at the same time 7.x is still far from being considered "production ready" (and the "RC" label clearly says that) and its development not that visibly fast. So, we need to wait unknown time and work as free beta-testers, devote hardware to this test (rumors are, someone bricked his/her devices during this full-of-fun testing time), and have no roadmap or trustable promises on release time or features to wait for.

You see, my interest is simple: software-defined networks (SDN) are a trend nowadays, and I'd like to use ZeroTier as kind of integrated crypto and network control solution all-in-one box but I'd better find another hardware vendor to deliver that.

Sometimes I think what if MT opened a project on Kickstarter to rise funds to develop 7.x to long-term state? I think I can donate some money, and believe some MT fans can do the same. Today the only significant 7.x development step was, mind that, when then went from old naming scheme to new one - but in fact it was only a scheme change, not got leap in development itself.

For the production readiness, you are right. For the rest you couldn’t be more wrong. New kernel, new routing engine, new read the rest on this forum.

About playing beta tester, that is a choice!

So, it seems I was a bit skeptical regarding upgrading couple of rb4011 from 6.48.4 to 6.49. Guess I was right - according to some reports, at least couple of rb4011 lose temp info (it displays 0K or -273°C). Hope they will address issues sooner than that dns mem leak bug.