v6.33rc release candidate (final testing)

What’s new in 6.33rc1 (2015-Sep-02 15:45):

*) ipsec - fix sockaddress buffer size on id generation for ipv6 address;
*) userman - refresh active sessions/users view dynamically;
*) package - added version tag and show everywhere alongside of version number;
*) rb493G - fixed reboot loop (introduced in 6.32);
*) wlan - improve single connection TCP performance for ac chipset with cm2 package;
*) defconf - fix default configuration for routers without wireless package;
*) mesh - fix router lock-up when interface is added/removed;
*) RB911/912 - fixed lock-up (introduced in 6.32).

*) wlan - improve single connection TCP performance for ac chipset with cm2 package;

Sounds nice!

Will 6.32 return in a fixed form or do we have to wait for 6.33 to become stable?

we will have a fixed one return as well, yes

I can confirm that my lab RB912UAG-2HPnD (broken by 6.32rc6 & 6.32 final) now has booted correctly on 6.33rc2 and I can winbox in it

Can I ask as a general rule what (basic) tests are done before releasing a RC ?
I mean a RC having reboot loop and lock-up ?
or are new issues like above just added to release checklist before releasing the RC.

I am repetitively asking for publishing quality assurance policies that mikrotik works according to, but I haven’t got any answer to it. I wonder if you will get info how they test the rc…

If we got a general checklist to what is tested before release, then we can decide how much we want to deploy into
our test network and a short time later maybe a section of the live network (?) and how much more testing maybe required before full deployment, in short we are trying to conduct a risk assessment to updates.

Actually you should not deploy any new version until it is publicly proved and you have made all your necessary tests.

Very good. Does this help with mixed environment .ac AP and .n CPEs ?
Can you share some background information what is happening (packet reordering ? ack aggregation effect ?).

Is this an improves also using NV2 protocol?
Any details about percentage of improvment?

Usually in software development, there are 3 (or even more) different types of versions:

Beta → Unstable. New features are added in this phase
RC → Release Candidate. In this phase, no new features will be added. If testing is OK, it will be released
Stable → Released version

Most of the past releases, I would describe as “Beta”.

I have no idea how MikroTik can publish new RCs every day.

https://en.wikipedia.org/wiki/Software_release_life_cycle

Maybe this would make people happy:

Call rc Beta
Call Current RC
Call Bugfix Stable

It is not about how we understand the naming. It is about our approach to the mikrotik software development life cycle. I would like to see newly released version that none will download and try just because none will want to risk the consequences.

And call RouterOS Beta RouterOS Alpha

Surely if you want to use the word ‘beta’ then ‘RC’ should be ‘alpha’?

Personally I think people should stop b*tching about the new release schedule, be greatful on the fact it’s better than the old system, and give it a few versions to bed in before they complain.

We have managed to reproduce issue with mangle rules disappearing. It will be fixed in upcoming 6.33rc versions.

As for rc version testing - Release Candidate is more likely to a nightly build. We will not do intensive testing before publishing these, only quick check if upgrade can be done and if most features work fine.

You are not forced to upgrade your router to rc version. Versions are released for those who has some problems on router and fix or possible fix is included in these packages. Basically rc is released for those who actually need it and for those who wants to do some testing and give us feedback.

6.33rc3 is released.

New fixes:
*) RB532/RB564 - fixed no link after ethernet disable/enable;
*) firewall - do not lose firewall mangle rules on start-up (introduced in 6.32);
*) tile - fixed occasional deadlock on module unload;

Is this related to [Ticket#2015073166000411] Lost capsman confing when upgrading ?

juanvi - No, it is not related. This fix is for SFP module unload.

Strods,

What will happen to 6.32 The fixed version that’s planned, now will the mangle rules being dropped in 6.32 will this be fixed in 6.32 fixed?

Dropping mangle rules on a reboot is pretty serious issue to leave unresolved to 6.33 unless its planned to fasttrack 6.33