I know over the years the question of whether or not to upgrade ROS comes up regularly.
Is there a consensus (LOL!!!) on when, if, and how to upgrade?
For example, I have a number of different model – hEX, ax2, ax3, ac, wap, cube, rb5009, etc.
Almost all are running 7.17.2 without problems (as long as I keep my fingers away from their configs and accept a couple of wifi quirkinesses/weaknesses).
Should I upgrade to 7.18.x?
If someone were running an earlier version (7.16.x, 7.15.x) would they get a different recommendation?
Do we hold by the “if it’s working, don’t touch it” approach?
Or something more abstract such as “the latest version fixes problems that we were not aware we had.”
One needs to look at the changes between updates.
For example 7.18.2 just came out and it has very few changes at all and most of them are for niche needs … the only one that may have broad applicability is improved stability of wifi interfaces, vague statements like this are not appreciated but one can at least surmize there is no wifi bug, just some sort of minor cleanup, and no admission if its from changes they made the last update or some outstanding issue never resolved before. crappy communications.
Go through 7.18.1 and 7.18 to see if there is anything that may improved your experience and functionality.
Typically its best to ensure all the routers talking are on the same Version, and ensure the embedded routerboard OS also matches the routerOS.
As @anav already hinted: have a look at new features, introduced since your running ROS version. If your current setup is running fine ,then you don’t really have to look at bug-fixes since any bugs that might be in ROS you’re running are not biting you.
The problem is that ROS v7 is still under active development, lots of new features are added … which inevitably means new bugs. And you never know what might bite you.
For me personally, I’m keeping a few sites on v6 (doing upgrade now and then just because) … and I have (my main ) site on v7. I went to 7.18.1 (from 7.17 or 7.17.1) for two reasons: IPv6 fasttrack and wifi/steering property 2g-probe-delay . Because these two features make my life much better. I might stay at 7.18.1 for a while now (none of change log entries for 7.18.2 seem to target my setup and quite likely 7.19 won’t bring anything new interesting to me). And my setup runs quite fine on 7.18.1, so no need to upgrade just because.
For devices which you don’t want to fuss around with, test things, possibly netinstall, a usual recommendation is to avoid using “bare” release versions (because they usually contain improvements/fixes that sometimes break things) and use the latest “point release”, which currently would be 7.18.2,
On how frequently to upgrade, I think the device-mode “allowed versions” parameter gives a bit of a clue. If I remember correctly, it’s currently somewhere around the 7.13 line of versions. The latest in that tree was released about a year ago. Allowing downgrades to that version would indicate to me that Mikrotik is not aware of any serious vulnerability. (Maybe it is because of the change in wireless packages…) So if things work stable for you and no security issues are published (or they don’t affect you,) I would suggest upgrading maybe once or twice a year.
If you browse the forum, there is some serious criticism about Mikrotik’s approach to software stability. This does not exactly line up with my experience. For me on a production system only one stable version caused a problem with a functionality that I used. (There was a bug related to sstp, but it was many years ago.) But as always, have a plan: make a backup/export, test thoroughly and prepare yourself - and others - for downtime should things go wrong.
Sounds like the overriding approach is that unless there is a fix for a currently-biting bug, or a feature improvement or addition, there is no reason to upgrade.
This is a straightforward and easy to understand approach to upgrading that still leaves some wiggle room individualized analysis – I like it.
This is a much more in-depth and thorough approach. That is, it requires a deeper understanding of the nature of new releases and the history of past releases, as well as taking into consideration one’s preparedness to fuss.
For people who “live and breath” ROS daily (or almost daily), this is a great approach.
For people who prefer (and are able) to forget about it’s existance, a simpler approach might be better.
For people in between (like me), this gets filed at the bottom of the long list of things to learn about.
7.17.x 1) is too fresh ; 2) and deplete completely the 16MB free space after some time…
7.18 is worse for point 2) and is unstable (see 7.18.2 and is not finished…)
7.17 is no longer being developed or fixed, so if no new fixes come out it’s not because there aren’t problems…
because with each new version, there being no long-term one, they completely abandon the previous one.
I understand the 16MB limit being sufficient to advise against 7.17.x
I'm not clean on what "too fresh" means. I understand the literal meaning that it has not been very long since its release, but that hasn't prevent MT from moving on to 7.18 and later.
Are you saying it is not worth of a long term stable designation? Because of bugs?
I’ll be vague, as I was before the CVE for brute-forcing usernames came out publicly.
I wouldn’t use 7.17 even if they paid me, I believe MikroTik got it done with 7.18…
but with 16MB peripherals I wouldn’t use 7.18.2 either…
and anyway 7.18.2 is too fresh…
) site on v7. I went to 7.18.1 (from 7.17 or 7.17.1) for two reasons: IPv6 fasttrack and wifi/steering property 2g-probe-delay . Because these two features make my life much better. I might stay at 7.18.1 for a while now (none of change log entries for 7.18.2 seem to target my setup and quite likely 7.19 won’t bring anything new interesting to me). And my setup runs quite fine on 7.18.1, so no need to upgrade just because.