@David-TP 

I am running in Accesspoint mode. To much of my network will get broken if I disable this. 

Therefore I am not able to enable connection alerts /  notifications. 

I can provide you with the deco log file tho. 

Home Asssistant reports this (to avoid getting a to long list these are the last 8 hours): 

August 18, 2023
esp8266 became unavailable       14:17:11 - 27 minutes ago
esp8266 became online                14:16:19 - 28 minutes ago
esp8266 became unavailable        14:12:37 - 31 minutes ago
esp8266 became online                14:11:27 - 32 minutes ago
esp8266 became unavailable        14:04:10 - 40 minutes ago
esp8266 became online                 14:02:09 - 42 minutes ago
esp8266 became unavailable        13:59:26 - 44 minutes ago
esp8266 became online                13:58:34 - 1 hour ago
esp8266 became unavailable        11:55:31 - 3 hours ago
esp8266 became online                11:54:55 - 3 hours ago
esp8266 became unavailable        11:53:54 - 3 hours ago
esp8266 became online                11:15:07 - 3 hours ago
esp8266 became unavailable        11:08:11 - 4 hours ago
esp8266 became online                11:07:36 - 4 hours ago
esp8266 became unavailable        11:06:30 - 4 hours ago
esp8266 became online                10:58:57 - 4 hours ago
esp8266 became unavailable        10:58:44 - 4 hours ago
esp8266 became online                09:43:52 - 5 hours ago
esp8266 became unavailable        09:42:46 - 5 hours ago
esp8266 became online                09:36:47 - 5 hours ago
esp8266 became unavailable        09:33:54 - 5 hours ago
esp8266 became online                09:30:28 - 5 hours ago
esp8266 became unavailable        08:02:53 - 7 hours ago
esp8266 became online                07:51:53 - 7 hours ago
esp8266 became unavailable        07:50:58 - 7 hours ago
esp8266 became online                07:43:43 - 7 hours ago
esp8266 became unavailable        07:43:32 - 7 hours ago
esp8266 became online                07:37:57 - 7 hours ago
esp8266 became unavailable        07:36:24 - 7 hours ago
esp8266 became online                07:34:41 - 7 hours ago
esp8266 became unavailable        07:31:58 - 7 hours ago
esp8266 became online                07:31:21 - 7 hours ago
esp8266 became unavailable        07:29:10 - 7 hours ago
esp8266 became online                07:27:45 - 7 hours ago
esp8266 became unavailable        07:16:13 - 7 hours ago
esp8266 became online                07:14:59 - 7 hours ago
esp8266 became unavailable        06:57:50 - 8 hours ago
esp8266 became online                06:40:00 - 8 hours ago

  @Er1 Yeah, that's one of the unfortunate things about 1.7.1 You MAY be able to obtain stability, but at the cost of arguably mesh's most important features...

I have used 1.6.x for a very long time. It was always very stable. 

I regret upgrading to 1.7.1. 

I have three m5's arround the house. Ethernet backhaul. 

They run in Accesspoint mode. 

Since upgrading the deco's my IoT devices (also mainly ESP8266) drop connection 20 times a day. 

Sometimes they connect back within 90 seconds. Sometimes it takes 90 minutes. 

I also sometimes have difficulties using the app. When I open it, it is trying to connect to the network but with no luck. 

This also only happend after 1.7.1.

All my wired connections that dont depend on TP-Link are rock solid. 

I've disabled Beamforming. 

I've disabled eco mode. 

I rather not disable fast roaming. 

Im using a local DNS (Adguard Home), and that is using Google's DNS (8.8.8.8, 8.8.4.4)

Everything else is default. 

(running M5 v3 1.7.1 Build 20230608 Rel. 74023)

  @David-TP Hi David, the two devices I am using are:

- for the WDS bridge: TPlink TL-WA901ND. Hardware revision V5, firmware version 3.16.9 Build 201030 Rel.58094n. I am running it in Bridge with AP mode. 

- for the powerline connection I am using TPlink TL-WPA4220 at each end of the link (hardware revision WPA4220 v1 08C84015, firmware version 1.2.9 Build 160525 Rel.35241n)

The bridging seems to work absolutely fine as I can access the network and the fiber connection no worries all the way down at the YARD2 node.

I turned off beamforming last night as a test and the majority of web access seemed to vastly improved as a result. Today it looks like the timeout issue has reoccurred for atleast two computers which would access via wifi. I carried out a full reboot of the deco network and the speed returned briefly, not sure what this indicates but its an observation?

The two computers which are having the worst issues have been connected to the LAN via cat6 for the moment until this can be resolved as it is affecting work activities during the day etc... especially teams calls and general browsing.

Overall the whole network has around 25 clients, the majority of which are CCTV IP cameras connected via cat 6 (and a switch) to YARD1 or YARD2. The access is excellent to these both within the network and also via internet connection via 4G. Other devices (Sky Q and NAS drive) are working well within the confines of the LAN.

It is a shame as 1.6.5 worked very well and we had no issues, presumably we can regress the firmware back to 1.6.5 without any major headaches??

EDIT: Even one of the machines ive connected using cat6 is now having connection problems with web pages so not sure what is happening here?! This needs to be addressed as one would hope a firmware update would improve things and not the opposite? :o

  @David-TP 

I've managed to get the network stable again by:

-Disable beamforward

-Disable fast roaming

-Set DNS to 8.8.8.8 / 8.8.4.4

Followed by a full power cycle of every single unit. We've been OK for about 4 days now.

However, these are features that were previously enabled and working without issue on 1.6.5.

Further, as expressed by others, device recognition has been quite spotty. I'm just getting a bunch of "network device" hostnames as opposed to the actual devices. And interestingly, a bunch of the Decos have just reverted to saying M5 instead of the set custom name. So now I have to re-identify them all within the topology.

So, to elaborate, 1.6.5 was stable and excellent, for the most part. Transitioning to 1.7.1 introduced a sleu of performance issues, especially with Deco's connected via WIFI backhaul. Most are hardwired. But they too were susceptible.

Post reboot, everything would funciton as expected but within a few hours, anything connected to any deco outside of the main would slow to a crawl, have trouble maintaining a connection, and/or just say no internet. I even noticed on my phone, the SSID would completely disappear.

The app would report all as fine, but clearly, things were not. I really should have gotten a dump, but it was more important to just get things stabilized. Once our season slows down, I can do that. 

Perhaps there is a mem leak or a rogue process causing the issues? This only seems to be affecting our larger network here. I have several other small M5 nets with like 2-4 decos and 1.7.1 seems OK so far.

  @David-TP Since 1.7.1 I notice all my Espressive devices are missing from the connected devices list.
All are connected, I can access these and Home Assistant communicate right with them using Wi-Fi so these devices are not offline.

It is really anoying because for example if I want to check if roaming is disabled as I wish I cannot because this setting is not even showed for offline devices.

These are both ESP32 and ESP8266 devices.

 Hi All, just to add my experience of the latest firmware (1.7.1) to the mix:

Ive got 5x m5 units which cover our house and our yard in the following configuration:

This has been working great for a year or so now, I get near ~50mbit download and maxmimum (20mbit) upload when running ookla speedtest at YARD2 which is the furthest from the main router node. Our connection is fttp 100/20mbit and is very stable.

Behaviour now:

YARD1 and YARD2 have flashing red lights and the deco app rightly states they are offline, however all connections still work fine!!?? I am able to connect to the wifi from either unit and use the internet connection as before. The devices wirelessly connected to either YARD1 or YARD2 appear as ethernet connected devices at MAIN when I look at the devices list in the deco app. We have IP cameras covering several buldings and they work fine so connectivity seems to be no problem.

Ive noticed also that web browsing from some (not all) devices seems to experience regular timeouts when following links and speed for programs such as teams/zoom is also erratic. On the whole we still have a fairly decent set up but it is quite erratic for certain devices and ive noticed the regular timeouts for web access doesnt seem to be consistant (ie some sites are fast, some seem laggy as hell!). EDIT: this is across all nodes so devices connected to ANNEX and KITCHEN also experience laggy behaviour (intermittently).

Settings are as follows:

- Fast roaming disabled, this was done some time ago as certain devices (android phone) were losing wifi all the time. Turning this off made it solid.

- Beamforming enabled.

- DNS servers are set to googles 8.8.8.8 server (and the correct secondary 8.8.4.4). This is for IPv4 and the equivalent v6 version. 

- SIP ALG enabled (not sure if this is strictly needed but we do use Voip)

- UPNP is on but most devices have it disabled

Its been pretty good up until around a week or two ago which coincides (ish!) with the 1.7.1 update. Not yet convinced to do a firmware regression back to 1.6.5 but unless a new update it forthcoming soon I might just do it and go through the pain of reconfiguring everything.

Hope this helps?!?? :)

EDIT: Ive also noticed even before 1.7.1 the device lists in the deco android app are not particularly up to date nor useful as they show devices attached to nodes i KNOW are not. Ie it will sometimes show a device connected to ANNEX when I know it is physically too far from it to possibly be connected there, it should be shown as connected to KITCHEN etc...