What the hell is going on (after upgrade to v5.1)

Hello, everyone. I use rb433, rb493, rb450, rb411 based devices all with atheros 5213 wireless cards, (sparklan mostly WMIA123AG). After upgrade ing from 4.17 which worked fine and stable for me I decidec to upgrade to version 5.1. From the beggining everything seems to work fine, but after a while routerboards looses connection with each other. After some time link is backed up and everything for some time seems to be working just fine.

LINK between mikrotiks are (bridge) ----> station wds (using nv2 wireless protocol)

From my point of view the v5 is very very and very unstable.

Have anyone had similar issues with newest routeros release ?

There is a lot of issues with version 5.1 - please go through the forum. For example 100% CPU load for some commands, some problems with chipsets, …

Last stable and working version is 4.17, for 5.x wait for 5.2, where should be a lot of bugs fixed (based on information on the forum).

Absolutely correct! Huge fixes are expected

Over the last week upgraded rb122’s, 133C’s, 532’s, 532A’s, 333’s, 411’s (all versions), 433’s (all versions), 493, 711’s, rb600 and rb800.
In total 230+ units.
Most of them upgraded remotely (wireless) and most from either 4.16, 4.17, any 5.0rc version and some from older 4.xx version. Did not have one single issue!

rb133C’s and 122’s that became slow under 4.16 and 4.17 and some 5.xrc’s are definately performing much better under 5.1. I’ve put 133C’s on the shelf as ´back dated´ to give them now a new life again with 5.1!

This is the first time in 5 years working with ROS that an relative new upgrade worked that seemless for me!
My network is now very stable (normal routed network with 802.11a, or ´n´ links and most now also running with NV2.)

First units upgraded now more than two weeks ago in production environment and not seen a single issue since!
4.17 running P2MP networks that still had some disconnect issues became very stable under 5.1.

ROS5.1 (and 5.0) is definately much more stable than last 4.x versions if it comes down to the wireless and simple routing.

I’ve seen some issues with 5.x still around but these are more exotic situation with tunnelling protocols, authentication setups, 863 platform or other exotic hardware. But on the wireless I have not seen too many issues mentioned by the experiance users.
Maybe its because I use only rb’s with MT radio cards in them. I’ve got any radio card they ever produced in my boards and none gave me problems on the upgrade.
Any major upgrade I performed in the past always left me with some units failing. This is the first time not a single board failed on me in the upgrade!
I don’t know how ´exotic´ the cards are mentioned. It can be MT needs to take a look on its drivers.


Since 5.1 came in my life blood pressure went down! :smiley: :smiley:

Imho if you have issues with 5.1 tell me your setup. Maybe I can give you helpfull advices.

Well Wirelessrudy, im happy that your boards work fine.
But my 450Gs and 750Gs arent usable with v5.1.
And this even with the default configuration, so no special setup.

I’m not buying MT is putting a ROS out which is not even running in default config on one of their own boards…
Tell me what you are doing exactly and maybe this forum can help you..

I’ve been a newbee myself years ago and made many mistakes in these days (probably still do at times!) I really don’t want to be remembered for… :wink:

This forum is here to help, not for showing frustration when something is not working the way you want.
On this forum more questions are asked or issues mentioned because of lack of knowledge and experience than as a result of bugs in the software..

Sorry to correct you, Mikrotik did indeed confirm this bug.
1.) In short do a fresh 5.1 Install vie Netinstall on a 450G or 750G
2.) run /export on level
3.) CPU up to 100%, SSH dead, Suppout dead/nearly dead

For details please see the other threads on this topic.

EDIT:
What’s new in 5.2 (2011-Apr-26 16:09):

*) console - fixed problem with supout file generation and export that
appeared in version 5.1, it was causing console to enter busy loop
on some boards;

OK, you win.. and shame on MT :confused:

I was merely replying to the other authors of this topic. They have other issues it seems and just the mentioning 5.1 is crap needed some correction. In general for boards most small wisps use it is a very good ros. Previous ROS family releases (3.x, 4.x) needed many more versions to reach same level of stability as 5.x
If I look in the change log for 5.2 I see only 5 fixes. Together with the only 4 fixes in 5.1 this is an all time low record for MT for the first set of upgrades!

This topic was started with the mentioning of wireless issues. No wireless issues are fixed in 5.2 so they either aren’t there or not fixed yet.
I don’t see nothing in 5.2 that would make me upgrade my network to it. 5.1 is stable and fine and doing the job it is supposed to do.

I second your oppinion that many issues might come from misconfiguratrion.
But if I learned one thing, then be carefull with new MT releases.
I really like MT but they are not any better than any other vendor today, testing is done by the customers. :confused:
And trust me, I know what I’m talking about, MT is my hobby-vendor, at work its Cisco, Alcatel, Adva, RAD and Huawei(well thats another story).

I’m with timberwolf.. if you’re going to run the latest release make sure you TEST the functions you rely on before rolling it out to somewhere 20m above the ground or 2 hours drive away :slight_smile:

Just because Mikrotik has tested it doesn’t mean you can’t find some way to break it!

I am having WAN to LAN throughput issues with 5.2 on RB750G. I only get ~30Mbps of my 100Mbps connection.
No problem with 4.17

Well, this only is a partial truth. Hardly anybody can create same circumstances on his test bench as in the real enviroment. Special when it comes to wireless and routing these can be tried on the bench but that give no quarantee it works in the ´real´ environment with many attached units and very variable data streams etc. The real environment is so much more complicated that that is the real test environment.

So the sequence is:

  1. Make a simple test on the bench. It works? Ok go to 2
  2. Upgrade a board on the real network that is not very important and can be accessed locally in an easy way, that works, got to 3.
  3. Do an upgrade to a remote unit, that still is easy accessible in case things go wrong, if that works, proceed to 4.
  4. Start to upgrade a compleet network, in a way that if things go wrong you still can relative easy correct things.

Save the most important / difficult units for last.
Also make backups for each unit first and store it on local pc/laptop

And then, when you upgraded your whole network and it runs… don’t be supriced after two weeks units start crashing because obviously there is still a bug around, that only floats onto the surface in very specific condtions…

So, the real test is in the real environment.

It’s like a car; how do you know al the safety measurments really work? Driving it in the parking lot of the dealer is not telling you anything. Use the car in a real environment and when the system then really is put to the test hope it works…
That’s why the manufacturer is crashing cars against the wall, and still can’t guarantee it always performs like intended…