Registered here and following this topic because I’m in the same struggle as others here, having MLAG not working under 7.14.2. Nothing in the changelog of the future 7.15 brings me hope, and I cannot downgrade to 7.6 some of my devices since they were originally shipped with newer versions.
Of course. There are many other things out there that I could replace with way more expensive substitutions to finally enjoy the advertised features...
But we're here on a Mikrotik forum, discussing about a RouterOS feature that should just... work (and used to, apparently, in previous firmware versions).
So apart from your patronizing answer, I would loooove to buy a Juniper QFX just to have MLAG finally working. You're welcome to finance me.
There are wide gaps between, say, a home installation with just a switch and a SOHO router, a homelab with somewhat advanced features, and finally a Tier 4 datacenter. Obviously people discussing on this board are eager to achieve better reliability in their setups with a usable MLAG function, but I'm pretty sure they're all quite aware that they won't build a datacenter-like setup with non-expensive hardware.
So I might be wrong, but despite being very well aware that expensive hardware is more prone to deliver high-end features (having worked with Cisco and Juniper in the past, we can all say that they also have their bugs and software regressions, sometimes even for simple features), I'm really not sure that advising us to chose other expensive equipment on this Mikrotik forum is going to help much.
This one feature was the whole reason that I specified Mikrotik for a few specific deployments.
Since this has been broken, Mikrotik has built-in lots of new features…Created lots of new problems…And yet this KNOWN and reproducible issue has gone seemingly without attention for over a year.
7.7 was released on January 12th 2023.
At this point, they should just completely remove the feature from the product and stop advertising that they can do it.
i agree with most of your statements. i also wished MT could deliver a working MLAG solution, but it seems they don't
of course more expensive hardware will deliver features differently but as i "advertized" switches from FS i was just trying to give options to working MLAG hardware other than juniper, cisco, extreme, etc. which most of the time are starting at a 4-digit price tag. a pity.
so FS might be a reasonable option for some people and those people trying to achieve MLAG setups are even willing to spend "a little more" for that MLAG reliability/benefit and have never heard of e.g. FS
haven't known that brand 3 years before and per today managing about 45 of those devices with little to no hurdles.
that's why i mentioned them. not to blindly point away from MT (been with MT for >10 years)
@spippan: Regarding FS, what do you think of their own FSOS compared to Mikrotik ROS or any kind of ONIE? Is there a big difference in cold boot time between them?
on e.g. N8560 or S5850 the CLI syntax is very cisco-like
boot time on a S5850 concludes about 3-5mins
boot time on a N8560 concludes about 5-8mins
never exactly measured them, but far from a reboot of, let’s say, a CCR2004-1G-12S+2XS (which is out of context, because there is no switch chip in it, sry, but just for reboot times)
Thanks for pointing me in the right direction guys. I got a pair of CRS326’s for the express purpose of running MLAG and even with 7.14.3 it completely breaks STP and then downs the entire network. If you disable MLAG, STP still doesn’t work until you reboot the switches. After that it correctly puts ports in alternate or backup mode as it should. These were going to be a top-of-rack pair in a co-lo rack…but even I’m not that masochistic!
Same here, have four 2116 running MLAG on 7.11
Ran the upgrade to 7.12.1. First pair went okay, but with the second pair I got burnt.
They made a loop in the network, lost contact with the local unit on the secondary link until after power down/up.
Eventually I managed downgrade all back to 7.11 but lost contact with the remote on the second link until after power down/up.
Ran a config compare between the one on 7.11 and 7.12.1 and did not find config changes.
Not sure MT broke MLAG beyond repair, but given the road to deployment and now the 1st upgrade (minor upgrade i dare say), it feels far from stable/reliable to us.
Not really I dare say. I have a pair of 326 as distribution switches on 7.15.1 running MLAG to the edge switches and also to the core. Recently I rebooted the primary and completely lost both. No more traffic to the core. Could not even access either one via MAC address, despite links were up. Had to go onsite. First I did a power reset on the secondary (the one that has not been rebooted before). And guess what, both units are running fine since then.
Conclusion: Once it works, don’t touch it anymore. That’s far from reliable and I will have to go back to Spanning Tree redundancy. Very sad.
Has support responded to anyone about these MLAG? Is there a bug tracker somewhere?
The reason I went down the road with Mikrotik was because of MLAG support on the CRS326’s. I now like MT for many other reasons but having flaky MLAG is going to cause me the re-think what I want to do. This is just for home setup but I’m trying to build HA everywhere as an exercise and MLAG is a key part of the puzzle. I even bought extra network cards for all my servers with this in mind. I guess I should have tried it first
Even if the same name FSOS , different by platform ( rebranded ).
The S series uses the CENTEC ASIC with the original OS ( they just renamed it, the versioning is the same as the manufacturer's ), matured platform.
Depends on the model, most of the S-series are using Broadcom nowadays (which you will see already on the product page which model uses which switchchip vendor). Some use Realtek and I think they might have some even older left in production which is using a chinese ASIC.
Except for their own gear with FSOS they also resell devices from Nvidia with Cumulus Linux as NOS and regular whiteboxes that comes preloaded and with a license from PicOS as NOS.
Yea…the only reason I bought Mikrotik was because of MLAG on the CRS518-16XS-2XQ.
I’ve had my share of issues during reboots, but most of the time my setup has recovered without manual intervention - but nevertheless I would never have bought those switches if they had not been advertised supporting MLAG
But except for those fun MLAG outages I’ve been happy with the rest of the features (bought two CRS518 switches and two CCR2004 routers for a small DC setup) for the money.