ESP8266 Community Forum

Hi, all. I'm looking into my ESPs not responding after a short time. It appears that the ESP is disconnecting from the AP / router after about 5 mins of inactivity even if there is a long time left on the lease. Below are the edited entries from my routers log.


It crossed my mind that a lease of 1440 mins (1 day) might be to long for the ESP to handle if this was being converted into micros (a bit like the maximum length of deep sleep is 71 mins) and my be being truncated to 32 bits. If this was the case and you took the 32 LSB from 1440 x 60 x 1000 x 1000, that equates to 8 mins. That sort of works if the ESP tries to renew the lease at 50% of its duration, tries and fails then disconnects. So I thought if I reduce the duration of the lease issued to 60 mins then this would fall within a 32 bit value, this however doesn't seem to make any difference.


Any thoughts as the log is not showing any activity from the ESP just a disconnection?

Fixed IP issued by the routers DHCP.
Mostly this code https://github.com/forkineye/E131 using the unicast option.

Could it be the WiFi (not DHCP) configuration on the router? I had one that would periodically renew the WPA keys, and some devices would disconnect.

Hi, @eduperez an interesting thought. I need to look more into this.
I've tried a different code from the same auther thet uses the E1.31.h file and it's been connected now for over 2 hours. Now I have 2 bits of code to compare. Unfortunatly the second one doesn't do what I need - isn't that always the case.
https://github.com/forkineye/ESPixelStick

Update: As far as I can see all the important bits of code are identical. So the issue must be somewere else.

Hi all, good news. I’ve part solved the problem (it’s taken some hours though) and there are still some big questions at the end of the post.

Quick answer:

If the ESP connected through the access point it was dropped from the DHCP list after 5 mins regardless of the duration of the lease. If it connected through the router it was maintained on the DHCP list and lease times where honoured.

Longer Answer:

My network hardware; I have a router (1) with cable internet. This then connects via CAT5 from a 1000 Mbit/s socket to a second router (2) which has had the DHCP disabled (all settings point at the first router), so it is basically an access point. The SSID and Password are the same on both routers.

Why I didn’t spot this; when the ESP’s connected they serial printed out their status (IP and what they were connecting to) as both the router and access point had the same SSID I was unable to discern that they were using different routes. It was only when I looked at the schematic type view on my router that I noticed that one ESP had a wired connection (via the second router / AP) and the other had a wireless connection (directly from the router). The logs that I had spent a lot of time looking at gave no clue as to the route of the connection.

What next; I turned off the second router / AP and power cycled the ESPs so both would be forced to connect to the router (1). Both have been up now for 5 hours despite the fact that I switch the second router / AP back on after an hour, it will be interesting to see what happens when the leases are up for first renewal attempt after 12 hours.

The questions:

Why if connected via this router / AP is it removed from the DHCP list after 5 mins?
I’ve done a lot of reading and a little knowledge is dangerous, so my thoughts – is something like this possible?
Asuming that the infomation below is correct (ok its windows specific but I’m guessing it’s not far off how things generally work).

Based on the fact that the ESP is dropped from the DHCP list after 5 mins. Could it be that the ‘DHCP acknowledge’ or ‘DHCP Nak’ is somehow blocked by the AP, so the ESP believes it has been unsuccessful and waits 5 mins before defaulting back to the request IP 0.0.0.0 and attempting to request a lease again? In the meantime the DHCP server has sent a ‘DHCP acknowledge’ so the IP is reserved and any traffic is processed as normal, the ESP having been given an IP even though it has not seen an ‘DHCP acknowledge’ processes any traffic on that IP and Subnet until the 5 min time out occurs. At this point it requests a new lease on the default subnet 255.255.255.255.
This is where I think it gets all messy. The router believes it has issued a lease to a wired device on a particular subnet and then receives a request from both a wireless and a wired device for a new lease with the same MAC on a different subnet and isn’t quite sure what to do so it just ends the current lease and ignores both.

It would be nice to get to the bottom of this before I get bored with the fact that I can make work anyway as quite a few people seem to have had problems with this.

All comments welcome.

Update: Both ESPs renewed their lease overnight (after 12 hours 50% of lease time) and are still connected to the Router (1). It's looking very much as if the problem is caused by the second router / AP (2).