V7.20.2 [stable] is released!

Announcements
42

I upgraded a hap lite with no problems.

43

This makes no sense. “Our train has a capacity of 2.000 people. It only has one door. When a lot of people want to enter or exit the train, long queues are formed in front of the door.” You don’t solve this by reducing the capacity of the train, you add another door! What about all people who don’t use PPPoE but now have to suffer a reduced connection tracking table size?

44

Confirmed, on 3rd attempt the update was successful

45

The connection tracking table size should be a settable parameter, as it is in Linux. It could be as number of entries with a displayed maximum or as a percentage of memory to be used for it.

46

i tested 7.20.2 the same problem with pppoe server over vlan, it’s not fixed.

SUP-201408
47

Only today in quiet hours I was able to generate the supout files from 7.19.6, 7.20.1 and 7.20.2.

It's not nat related. SUP-202674

48

Where can I find the upgrade path. Currently running 7.14.3. I want to upgrade to 7.20.2

49

Just upgrade from 7.14.3 to 7.20.2 in one step. It always offers the suitable next version when you select “check for updates” in system→packages which you can then download and install.

Only 7.12 would be an intermediate step when you had a v7 before that.

50

Yep - connection tracking disabled. Tik shared a ‘daily’ build with me where it is apparently fixed, but I haven’t tested it yet. Frankly, not to keen to even look at 7.20 tree due to the BGP blackhole thing. So stuck between a rock and a hard place here at the moment.

51

Since upgrading to 7.20.2 I’m seeing ethernet rates being negotiated that are below what they should be, i. e. 1 Gbps link being negotiated as 100 Mbps. Seems to be an intermittent issue as rebooting apparently fixes it temporarily. This is on a CRS328-24p.

52

Sure this is caused by this relase, @Smoerrebroed? Have you tried chaning cable?

53

Yes, tried a different cable and a different port. I don’t see this with all my devices though, so perhaps it’s a combination between the CRS328 and my client device (Lenovo Thunderbolt 4 Dock).

54

And the computer connected to that dock is active when this happens?

It's known (perhaps not very well) that if computer is in a sleep mode, then network interface down-switches to lowest supported speed (to save some power) ... which used to be 10BaseT (half duplex) but with Gbps (or multi-Gig) interfaces it might be 100BaseTX. After device wakes up, it should re-negotiate to highest speed supported by both link parties.

55

It is known, and it’s the same behavior with several different machines connected (to the dock). My CRS312 doesn’t seem to be affected though.

56

Hello

push-continuation received from OpenVPN server in RouterOS OpenVPN Client causes client to freeeeze and … do nothing after geting second line of PUSH ROUTES

It seem’s to working in 7.19.6 last times.

57

Dual ISP load balancing is not working anymore 7.20.2

58

Ideally, you should finally implement multicore for the PPPoE process. And that will solve these problems for those of us who use PPPoE with or without VLANs. MikroTik, pause the changes for a moment and fix PPPoE once and for all. We'd really appreciate it.

59

In our case, using version 7.19.6, after X clients using NAT for the cloud management system, the clients are disconnected. And the router user is also lost, remaining with the default user.

60

More issues with V7.20.2

CCR2004-16G-2S+ software version 7.20rc5 was working well. Upgraded to 7.20.2 now I'm having issues with SFP modules. Upon reboot the 10G RJ45 SFP module will not work. If you unplug the patch cable and reconnect it, it works. If you disable and enable the SFP it will start working.

I had to write a script and schedule it on boot up to disable and enable the SFP to get this to work.
Tested beta 7.21beta3 its still broken, but different. The SFP lights will blink, like its trying to reboot, but never works. Unplugging the patch cable and reconnecting fixes this issue too.

The fix under schedule, startime = startup
On Event = delay 5; /interface/set sfp-sfpplus1 disable=yes; delay 5; /interface/set sfp-sfpplus1 disable=no

61

Ok, So looking for a bit of info if anyone is able/willing.

Primarily arround “firewall - reduce maximum connection tracking entry count;”

1 - What has it been reduced FROM - and when was the higher limit implimented - I see several complaining it’s @ 1048576 - but I was under the impression that was the limit for a long time.

2 - Why was it reduced - i’d like to be able to confirm if the issue was something I was ever going to hit - in say a Hotspot environment.

3 - the Other “related” PPPoE issues - can anyone confirm that these issues do not happen when connection tracking is not implimented.

I am concerned as i’ll be looking to rebuild our PPPoE Server soon. I currently run BGP - no NAT - PPPoE Server, so do not have any contrack issues (except when I want to track them… lol).

I do however need to impliment DHCP in my next build, and was considering adding in CGNAT at this time - which would 100% be for DHCP - but pool “May” be shared with PPPoE (for non static addresses).

We have several thousand IPs currently - so have a big of space to live without NAT for now - but looking to pack things in for the future, and not re-build until upgrading to 400Gb.

Also - just as a point of interest - how many routers do you all have connected that hit that 1M connections limit?

Cheers for anyone who responds.