[SendToHttp] Do not set HTTP authentication when no user/pass set #4205

TD-er · 2022-08-16T17:08:34Z

A number of setups may report HTP error 401 since the 20220809 build.
This fix should not add an auth header if no user/pass is set.

Fix wait for acknowledgement flag / timeout (Generic and Advanced HTTP controllers timeout despite Ignore Acknowledgement option is set #4208)
Fix HTTP 401 error code where old builds worked fine (Latest Release ESP_Easy_mega_20220809_normal_ESP8266_4M1M don't send HTTP in rules. #4203)

Fixes: #4203
Fixes: #4208

A number of setups may report HTP error 401 since the 20220809 build. This fix should not add an auth header if no user/pass is set.

tonhuisman · 2022-08-16T18:09:53Z

This should fix #4203 , @fly74 can you check if it now again works as expected, using the Github Actions build?

fly74 · 2022-08-16T18:12:04Z

@tonhuisman Sure but what is a action build?

tonhuisman · 2022-08-16T18:51:08Z

On top you see a tab called Actions, there you see a list of recently run Actions, one of them being the build of this Pull Request:

Clicking the title will bring you to the overview page for that Action, that has a link to the results:

In that Binaries zip file, you will find a (bunch of other) zip file with the build you normally use. Extract that and you will find the .bin files, in a subfolder.

fly74 · 2022-08-16T19:21:20Z

Testet with ESP_Easy_mega_20220816_normal_ESP8266_4M1M, same problem, not fixed.

TD-er · 2022-08-16T20:06:38Z

Can you give an example of what you're trying to send and perhaps to what kind of webserver?

fly74 · 2022-08-17T04:51:58Z

@TD-er

242670 : Info : EVENT: MS#State=1
242678 : Debug : timeStringToSeconds: "1" --> invalid
242678 : Debug : timeStringToSeconds: "1" --> invalid
242681 : Debug : conditionMatch: "1=1" --> "1" = "1" --> true (1 = 1)
242681 : Debug : conditionMatchExtended: true "1=1"
242682 : Debug : Lev.1: [if true]=true
242684 : Info : ACT : SendToHTTP,esphost2,80,"control?cmd=event,pon"
242686 : Debug : Command: SendToHTTP
242687 : Debug : SendToHTTP,esphost2,80,"control?cmd=event,pon"
242688 : Debug : Par1: 846230272 Par2: 80 Par3: 1073710592 Par4: 0 Par5: 0
242691 : Debug : SendToHTTP: Host: esphost2 port: 80 path: control?cmd=event,pon
242738 : Info : HTTP : Start Digest Authorization for esphost2
242806 : Error : HTTP : SendToHTTP esphost2 GET... failed HTTP code: 401
242812 : Info : ACT : SendToHTTP,esphost1,80,"control?cmd=event,pon"
242815 : Debug : Command: SendToHTTP
242815 : Debug : SendToHTTP,esphost1,80,"control?cmd=event,pon"
242816 : Debug : Par1: 829453056 Par2: 80 Par3: 1073709568 Par4: 0 Par5: 0
242819 : Debug : WiFi : Set TX power to 0dBm sensitivity: -69dBm RSSI: -47dBm
242820 : Debug : SendToHTTP: Host: esphost1 port: 80 path: control?cmd=event,pon
247062 : Error : HTTP : SendToHTTP esphost1 GET... failed HTTP code: -1 connection failed
247069 : Debug : Lev.1: [else]=false
247075 : Debug : EVENT: MS#State=1 Processing time:4405 milliSeconds
247079 : Info : EVENT: http#esphost2=401
247081 : Debug : EVENT: http#esphost2=401 Processing time:2 milliSeconds
247083 : Debug : LoopStats: shortestLoop: 57 longestLoop: 4568311 avgLoopDuration: 213.17 loopCounterMax: 526315 loopCounterLast: 152269

I hope I used the action build correctly, the esphost1host is not running and the connection failure for this is correct.

ghtester · 2022-08-17T14:13:07Z

FYI I have just quickly tested the Generic HTTP & Generic HTTP Advanced controllers on latest Custom ESP firmware (compiled yesterday without PR4205) with Port Listener app. as a target.
The strange thing is that despite I have the Check Reply: option set to Ignore Acknowledgement, I am getting the connection errors due to timeout (which is set to 200ms and the connection is local):

5503126: HTTP : C008 192.168.44.34 GET... failed HTTP code: -11 read Timeout
85503127: HTTP : C008 connection failed (521/0)
85503129: EVENT: http#192.168.44.34=-11
85503531: HTTP : C008 192.168.44.34 GET... failed HTTP code: -11 read Timeout
85503532: HTTP : C008 connection failed (522/0)
85503536: EVENT: http#192.168.44.34=-11
85503813: HTTP : C008 192.168.44.34 GET... failed HTTP code: -11 read Timeout
85503814: HTTP : C008 connection failed (523/0)
85503817: EVENT: http#192.168.44.34=-11

And the target accepted the connection repeatedly (as it is just listener, without any reply to initiator):

Client connected
GET /demo.php?name=ESP_04&task=Voltage&valuename=Analog&value=39 HTTP/1.1
Host: 192.168.44.34
User-Agent: ESP Easy/20116/Aug 16 2022 14:06:46
Accept-Encoding: identity;q=1,chunked;q=0.1,*;q=0
Authorization: Basic Og==
Connection: keep-alive
Content-Length: 0

Client disconnected

…ttp_401_error

Still to do: look into why not all HTTP requests seem to arrive at the other node when sending at short intervals (less than 3 sec)

TD-er · 2022-08-23T21:12:10Z

I also fixed some issue where (especially on ESP32) a second call to some HTTP server may fail until some (long) timeout or another call to a different host was made.
However there is still something I need to look into as a number of HTTP calls in short intervals may still lead to missed calls even though the call will be registered as a success (or at least the same HTTP code as the previous call)

@ghtester can you test this PR?

ghtester · 2022-08-24T03:29:16Z

Thanks for your work, as already reported in #4208 thread, unfortunately I still see the Timeout issues.

WiFiClient has a setTimeout function which expects seconds. The parent class, Client, does expect msec. So by casting to its parent and then calling setTimeout, we can again set the timeout in msec.

As discussed here: letscontrolit#4208 (comment)

Otherwise it is set to 4000 msec. For servers that won't send an ack, you would otherwise always hit this 4000 msec timeout, which is way too long.

The set timeout is now considered a minimum timeout. If it isn't enough, it will dynamically rise until it is enough. Typical used timeout is 3x the last encountered duration, with a minimum set in the controller settings.

This setting is set for a reason, like the server may never reply/ack. If set, the duration will always match the set timeout and thus increase over time.

TD-er · 2022-08-24T10:40:30Z

For the controllers, I made the timeout dynamic (unless it is set to ignore ack).
The timeout will now be 3x the last actual duration, with the set timeout as minimum.
If it is set to ignore the server ack. this will obviously not apply.

Still thinking whether I should make this dynamic behavior optional.

[SendToHttp] Do not set HTTP authentication when no user/pass set

8b6e002

A number of setups may report HTP error 401 since the 20220809 build. This fix should not add an auth header if no user/pass is set.

TD-er added 6 commits August 21, 2022 09:49

Merge branch 'mega' into bugfix/sendToHttp_401_error

3ecad1f

Merge branch 'mega' into bugfix/sendToHttp_401_error

2b2d5ab

Merge branch 'mega' into bugfix/sendToHttp_401_error

9fb4940

Merge branch 'mega' into bugfix/sendToHttp_401_error

253b892

Merge remote-tracking branch 'letscontrolit/mega' into bugfix/sendToH…

b25f769

…ttp_401_error

[SendToHttp] Fix HTTP 401 and 404 error

b87db70

Still to do: look into why not all HTTP requests seem to arrive at the other node when sending at short intervals (less than 3 sec)

TD-er mentioned this pull request Aug 23, 2022

Generic and Advanced HTTP controllers timeout despite Ignore Acknowledgement option is set #4208

Closed

TD-er added 6 commits August 24, 2022 10:45

[Network Timeout] Apply workaround to set timeout in msec on ESP32

2c5d722

WiFiClient has a setTimeout function which expects seconds. The parent class, Client, does expect msec. So by casting to its parent and then calling setTimeout, we can again set the timeout in msec.

[Cleanup] Uncrustify _N001_Email.cpp

85643ec

[HTTP] Change read timeout into HTTP code 200 when "Ignore Ack" is set

84b3cf5

As discussed here: letscontrolit#4208 (comment)

[SendToHttp] Set timeout to 1000 msec when set to ignore ack

f542e15

Otherwise it is set to 4000 msec. For servers that won't send an ack, you would otherwise always hit this 4000 msec timeout, which is way too long.

[Controller] Make timeout dynamic, depending on actual results

bd98bdc

The set timeout is now considered a minimum timeout. If it isn't enough, it will dynamically rise until it is enough. Typical used timeout is 3x the last encountered duration, with a minimum set in the controller settings.

[Controller] Don't make timeout dynamic if ignore acknowledgement is set

00a7839

This setting is set for a reason, like the server may never reply/ack. If set, the duration will always match the set timeout and thus increase over time.

TD-er merged commit 275a9f1 into letscontrolit:mega Aug 25, 2022

TD-er deleted the bugfix/sendToHttp_401_error branch August 25, 2022 10:08

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[SendToHttp] Do not set HTTP authentication when no user/pass set #4205

[SendToHttp] Do not set HTTP authentication when no user/pass set #4205

TD-er commented Aug 16, 2022 •

edited

tonhuisman commented Aug 16, 2022

fly74 commented Aug 16, 2022 •

edited

tonhuisman commented Aug 16, 2022

fly74 commented Aug 16, 2022

TD-er commented Aug 16, 2022

fly74 commented Aug 17, 2022 •

edited

ghtester commented Aug 17, 2022

TD-er commented Aug 23, 2022 •

edited

ghtester commented Aug 24, 2022

TD-er commented Aug 24, 2022

[SendToHttp] Do not set HTTP authentication when no user/pass set #4205

[SendToHttp] Do not set HTTP authentication when no user/pass set #4205

Conversation

TD-er commented Aug 16, 2022 • edited

tonhuisman commented Aug 16, 2022

fly74 commented Aug 16, 2022 • edited

tonhuisman commented Aug 16, 2022

fly74 commented Aug 16, 2022

TD-er commented Aug 16, 2022

fly74 commented Aug 17, 2022 • edited

ghtester commented Aug 17, 2022

TD-er commented Aug 23, 2022 • edited

ghtester commented Aug 24, 2022

TD-er commented Aug 24, 2022

TD-er commented Aug 16, 2022 •

edited

fly74 commented Aug 16, 2022 •

edited

fly74 commented Aug 17, 2022 •

edited

TD-er commented Aug 23, 2022 •

edited