ROS 7.2 broke entire local network

RouterOS General
1

Hi,

I’m using several MT devices on local network:

1-CCR1009-7G-1C-1S - Main router
2-wAP G-5HacT2HnD_A - First AP
6-RBwAPG-5HacT2HnD.rif - Second AP
3-RBSXTR - LTE Modem - Backup internet
5-CRS305-1G-4S+ - 10G Switch
7-RBwAPR-2nD - Lora Gateway

Everything was working corectly on 6.49.5. I’ve made manual upgrade to ROS 7.2 all devices except main router ( CCR1009-7G-1C-1S+ ) and then main router starts behave abnormally. It starts loosing 50% packet on almost every port. So I’ve upgrade it to RoS 7.2 but it doesn’t fix the problem.

Problem gone when I’ve phisically disconnect almost every MT device from list ebove ( except 2-wAP G-5HacT2HnD_A ). After reconecting again, problem has returned.

What is more interesting, SFP+ port which was connected to QNAP 10G also behave strange. Please check attached video. You will see that value in field ‘Connector Type’ and OM4 disappear for a moment. For test I’ve also connect this port to another CRS305-1G-4S+ using MT DAC but it behave in the same way. Both value in fields disappiring for short time periodically.

I didn’t make support file from this situation, couse I thought that router has been damaged. Fortunetelly downgrade all devices to 6.49.5 restored proper netowrk working on 1-CCR1009-7G-1C-1S+

support file for all above MT devices was send to support@mikrotik.com. but from situation when everything works normally ( from 6.49.5 )

I have no idea what may couse such problems. I’m using also dude server on CCR1009-7G-1C-1S+

Is support file from problematic situation will be required, I will try to once again upgrade most devices to 7.2 and reproduce above situation.

https://drive.google.com/file/d/1DnUFZfsaVY9iPrRhuqeTk2EjThbOAvD6/view?usp=sharing

2

Sorry cannot be of assistance from a crappy picture of a computer screen.
I also dont understand your logic where the CCR has been working and is stable would some how magically be the problem.
The one thing I would not have done is change the firmware on the CCR until the rest of the devices are sorted out.

There are some syntax difference between 6 and 7 and differences in how IP routes with tables and route rules are created for example
and most likely there are many others and thus a copy of config between versions is not going to cut it for all settings.

Without the configs of the devices in question, too hard to make guesses.

3

Hey, I had a similar experience that I reported in the v7.2-topic: http://forum.mikrotik.com/t/v7-2-is-released/157082/1

Update from 7.1.5 to 7.2. results in massive packet drop, unresponsive router. The box is a CCR1009, I had dude on it previously, but migrated it to another box before update to v7. Switching to a second partition with 7.1.5 brought the previous running state back. I know that this is not very detailed, but as this is a production router I don’t have the time now to debug the problem. Maybe next week I will be able again to play a bit with 7.2 but it is interesting to see a similar/same problem on a CCR1009.

UPDATE: The problem was solved once I configured my CAKE queue types correctly. As long as I had CAKE in simple queue with bandwidth parameter set, I could trigger the packet drop/latency by enabling and disabling the simple queues.