Topology... why isn't this working anymore....
So I have been suffering with the absolutely nuts packet loss issued mentioned here:
https://community.eero.com/t/y4hj10g/what-have-you-done-insane-packet-loss
I have had eero nodes for over 5 years, if not longer. All in bridge mode.
I have had as many as 7 nodes connected top provided connectivity to the house, especially at the start of the pandemic when I was support 5 at home students.
About 9 months ago, it has gone to crap, and the stability of the network is all over the place. I will get a total loss of traffic on the wired and wireless network, sometimes just the wireless. The only thing that works to restore, is power cycling ALL the eero nodes.
I removed 2 nodes, because of concerns there were too many.
I updated 3 First Gen to 6e's ($700), no change.
I added the TPLink 2.5g switch to the mix and re-wired some runs, so each eero, has it's own home dedicated run, with nothing else attached to them via wire... no fix.
I am at a loss. In the images is my current topology, which should totally work. And did for YEARS.
The 2nd image is what eero support is suggesting I do, to resolve the problem.
And I am scratching my head on how it is supposed to fix anything, other then have ALL my traffic go through the one eero node.
I am NOT using any of the software feature that eero has added the last several years, I don't need it or want it. I just want the mesh WiFi access points, that eero did very well for every long.
I am nearing the point, that I am just going to scrap it all.. .and rebuild with Ubiquity or something else.
We rely on this network for use about everything in the house (24+ smart switches and devices, fire detectors, water sensors, cameras), Alexa, Google Home, streaming TVs, wireless devices, PC's, printers. yes, I am by far an above average home user, but this should and did work.
I have tested every home run to rule out a cable that has gone bad.
I have replaced some patch cables that "may" have been compromised.
None of that has changed anything.
The only thing that restores the network, is a power cycle of all 5 eero nodes, and rebooting one at a time, letting it power up.. and then the next, and then the next.
10 replies
-
Hi ebonovic
In order for the eero to function correctly there has to be one eero upstream of all other hardwired eero or coordinate communication. The reason for this lies in how packets can be transmitted from a device to the gateway across an eero mesh network. Whenever any particular eero needs to send data from point A to point B it will send it via the lowest cost method to the gateway/head eero. So it might send one on the 5.2 band to eero B (which will then get to the gateway), another 3 via an ethernet link and then the final 2 via the 5.2 via eero C (since B is busy). Of course if some of those intermediate eero might send the packet via who know how to the gateway. Normally ethernet links are chosen as those have 0 airtime cost (they don't add any busyness to the airwaves) but that is not a hard rule. if the ethernet is busy the eero could send the packet via wireless.
So, with the eero mesh the packets might arrive in any order at the gateway (so for 5 packets, it might arrive in order: 01432 instead of 01234). For wireless this is no problem, that is accounted for in wireless originated packets, the receiving device waits for all the packets to arrive and then reorders them into proper order. For wired communication it is a really big problem. When ethernet was originally designed way back in the day it was designed for everything to arrive in order, 01234. It has no provisions for things to arrive out of order. So you can see that with the eero mesh that could make wired packets useless. eero, however, has an algorithm specifically built to handle this called STAMP. It makes sure that all packets are sent upstream in the order they need to be (which is an over-simplification of it, but it illustrates what it does in this case). However, for STAMP to work, one eero must be upstream of all the others in order to be the gateway eero and to run that algorithm for all the eero. With your original setup, eero A might be the gateway and is trying to organize, but eero B, since it has equivalent wired access to the router as A, might send one of those packets to the router and poof, wired communication is broken.
I do have a change to the recommendation. Instead of using the topology in the second image of Edgemax->eero 6E->tp link switch->netgear switches and eero I would instead go
Edgemax->eero 6E and netgear switches and from eero 6E->tp link switch->other hardwired eero.
Since the eero are bridged and not in control of the whole network this should do. You will have one eero above the other hardwired eero, so that will fulfill the topology requirements without affecting the rest of the hardwired network.
https://support.eero.com/hc/en-us/articles/360000830546-Examples-of-common-network-topologies
-
Michael_eero_support
So here to report... no change.
I made the requested changes. Things were good for about 36 hours, then last night about 10pm the entire network (wireless and wired) locked up. I disconnected an Eero node, and things restored. Still have that node disconnected.
This morning, the entire network locked up again. I was at my desk, and unable to ping the gateway with packet losses, so it is not a WAN issue. I disconnected the gateway Eero (which in this new topography, disconnects all Eero's), and immediately restored connectivity. let it stay that way for about a minute, to watch the ping traffic... all good. The moment I reconnected the gateway Eero, packet losses.
I just restarted the gateway Eero, and for the moment things are stable.
But this is not a sustainable option for me. where do I go from here?
-
Michael_eero_support
I will send that email in just a moment.
As for a different eero as the gateway, no have not tried this.
This particular one that I am using now is one of the ones I purchased this summer, the 6E.
I purchased this set this summer, as I thought my issue was possibly related to the Gen 1's going bad.
So while it is always possible that this one is faulty, the problems existed before it's introduction, and after. So the chances are minor.
It is an option to attempt though.
Content aside
- 2 yrs agoLast active
- 10Replies
- 532Views
-
3
Following