Connected2.homeseer.com not working

[The Profit]

I wonder if the geofency app is caching an IP address? For example if it accesses connected2 and it returns the IP address for connected11 it then saves that IP. If we then switch to connected22 the app should detect this (based on the TTL timeout which is about 15 seconds) and switch to the other IP. If this is an app that you run, next time you see a connection issue restart that app and see if it then works. I know there are systems that are not honoring the TTL. Right now we are on connected22 but I see 13 systems connected to connected11. Those systems will run into issues.

I think you are correct that the Geofency app is cashing an IP address. I left the house this morning at 11:19am, using the connected2 link in the app. The change didn’t record in Homeseer that I had left, so because I don’t keep the Geofency app running/open in the background, I powered my phone off and back on. When I got home, Homeseer had received the change in the device.

When you move to a load balancing server will that possibly resolve these kind of issues?

Thank you for your help with this!


Sent from my iPhone using Tapatalk

[rjh]

Yes, when we go to load balancing we will redirect requests to the server your HS system is connected to. My plan is to turn that on next week.

I think you are correct that the Geofency app is cashing an IP address. I left the house this morning at 11:19am, using the connected2 link in the app. The change didn’t record in Homeseer that I had left, so because I don’t keep the Geofency app running/open in the background, I powered my phone off and back on. When I got home, Homeseer had received the change in the device.

When you move to a load balancing server will that possibly resolve these kind of issues?

Thank you for your help with this!


Sent from my iPhone using Tapatalk

[logman]

I wonder if the geofency app is caching an IP address? For example if it accesses connected2 and it returns the IP address for connected11 it then saves that IP. If we then switch to connected22 the app should detect this (based on the TTL timeout which is about 15 seconds) and switch to the other IP. .

You could set the TTL to 0, which would prevent caching by downstream DNS resolvers and force all queries to reach Traffic Manager for resolution.

--Barry

[mikee123]

There still seem to be issues with connected2. My tasker occasionally works again, but last night it didnt.

[The Profit]

There still seem to be issues with connected2. My tasker occasionally works again, but last night it didnt.

Something that Rich suggested that worked for me was to shut down the app on your phone and restart the app. If that doesn’t work, try restarting the phone.

The issue seems to be that the app on your phone is caching the first IP address that works and doesn’t update it when it changes. If the above suggestions don’t work then submit a ticket to support.


Sent from my iPad using Tapatalk

[SteveMSJ]

I wonder if the geofency app is caching an IP address? For example if it accesses connected2 and it returns the IP address for connected11 it then saves that IP. If we then switch to connected22 the app should detect this (based on the TTL timeout which is about 15 seconds) and switch to the other IP. If this is an app that you run, next time you see a connection issue restart that app and see if it then works. I know there are systems that are not honoring the TTL. Right now we are on connected22 but I see 13 systems connected to connected11. Those systems will run into issues.

rjh

Connected2 is again not working for Geofency webhooks. I get the same using Chrome on a PC before it redirects to the Log In page, see snapshot below for chrome on PC and Geofency on iPhone:





Arrived home after the holidays to a cold house

I'll email support.

Steve

[rjh]

See if its ok now.

The service switched to the backup but now its back on the primary. This should not cause any issues, but I guess whatever service you are using is not honoring the TTL on the DNS and is keeping the old IP. We are implementing a re-direct to assist with this type of issue, but if the server its trying to connect to is actually down, a redirect is not going to help.

rjh

Connected2 is again not working for Geofency webhooks. I get the same using Chrome on a PC before it redirects to the Log In page, see snapshot below for chrome on PC and Geofency on iPhone:





Arrived home after the holidays to a cold house

I'll email support.

Steve

[SteveMSJ]

See if its ok now.

The service switched to the backup but now its back on the primary. This should not cause any issues, but I guess whatever service you are using is not honoring the TTL on the DNS and is keeping the old IP. We are implementing a re-direct to assist with this type of issue, but if the server its trying to connect to is actually down, a redirect is not going to help.

Yes it is working again now thanks.

It’s a bit of an issue when this happens as I’m pretty sure a large number of people use Geofency and PHLocation with HS3.
Steve

[zwolfpack]

... This should not cause any issues, but I guess whatever service you are using is not honoring the TTL on the DNS and is keeping the old IP. ...

I had an issue last night with GeoFency last night as well. The issue is not with GeoFency - below present indications that the issue lies within the HS infrastructure.

Returning home at Jan-01 1:50 AM PST, noted that Geofency based events did not run as expected. These had run OK when leaving Dec-31 8:35 PM PST.

In HS3 log, searching for "MyHS" noted a single interruption in the MyHS service, at 9:47 PM PST

Dec-31 21:47:46         Info    System connected to MyHS Service successfully with license ID xxxxx.
Dec-31 21:47:45         Info    System connected to MyHS Service, waiting for acknowledge...
Dec-31 21:47:31         Warning    Have not heard from MyHS in over 3 minutes, disconnecting for re-connection.

On my HS3 server (Hometroller S6, Windows 7, HS3 3.0.0.548), at approx. 2AM PST noted the MyHS backchannel connection:

C:\Users\homeseer>netstat -tn | findstr 10300
  TCP    192.168.1.103:51832    [COLOR=#FF0000]45.77.216.233[/COLOR]:10300    ESTABLISHED     InHost

Which corresponds to connected33.homeseer.com.

However DNS (on that system and elsewhere) were showing connected2.homeseer.com was resolving to connected22.homeseer.com

C:\Users\homeseer>nslookup connected2.homeseer.com
Server:  UnKnown
Address:  192.168.1.1

Non-authoritative answer:
Name:    [COLOR=#0000FF]connected22.homeseer.com[/COLOR]
Address:  [COLOR=#FF0000]45.76.21.192[/COLOR]
Aliases:  connected2.homeseer.com
          connected2.trafficmanager.net

So it appears my issue was due to my HS3 server backchannel connected to the "wrong" server.

Checking today, in HS3 log see

Jan-01 10:01:37         Info    System connected to MyHS Service successfully with license ID xxxxxx.
Jan-01 10:01:36         Info    System connected to MyHS Service, waiting for acknowledge...

And the backchannel IP now matches connected2/connected22:

C:\Users\homeseer>netstat -tn | findstr 10300
  TCP    192.168.1.103:63468    [COLOR=#FF0000]45.76.21.192[/COLOR]:10300     ESTABLISHED     InHost

So whatever you did at 10AM PST (11 minutes prior to your post #24 above) apparently reset my MyHS backchannel, and is now working.

So I'd question what mechanism you have in place to ensure that existing backchannel connections are reset in the case where they don't match the currently active frontchannel.

[integlikewhoa]

Seems to be temporarily working again, but let it be known not only is geofencing effected so is the use of Imperihome using MyHS.

[Corvl1]

Chiming in here as well , due to problems with imperihome-homeseer , connecting gives an error :"invalid response from AP"

Thanks,
Cor

[rjh]

Imperihome issue is known and once the redirect is implemented it should then be ok. Since I don't know how the geo thing is working I cannot say if the redirect will work for that.

Chiming in here as well , due to problems with imperihome-homeseer , connecting gives an error :"invalid response from AP"

Thanks,
Cor

[Corvl1]

rjh ; thanks for your reply. Do you have an estimate for this redirect?

Thanks,
Cor

[zwolfpack]

Imperihome issue is known and once the redirect is implemented it should then be ok. Since I don't know how the geo thing is working I cannot say if the redirect will work for that.

The 'geo thing' works by posting to https://connected2.homeseer.com/phl2api?dtype=GF, which it does correctly.

Does the 'redirect thing' address the situation where the HS3 application remains connected to the non-active connectedxx.homeseer.com:10300, as shown in post #26 above?

[zwolfpack]

MyHS switched over to an alternate server today @12:38 PST.

Jan-03 12:38:44

Info

System connected to MyHS Service successfully with license ID xxxxx.

Jan-03 12:38:44

Info

System connected to MyHS Service, waiting for acknowledge...

Jan-03 12:38:33

Warning

Have not heard from MyHS in over 3 minutes, disconnecting for re-connection.

connected2.homeseer.com is currently resolving to connected33.homeseer.com. Prior to that was resolving to connected22.homeseer.com.

Services are working properly, since the HS3 backend reconnect logged above was established to the 'primary', connected33.

Will continue to monitor. In the 12/31 evening failure, connected2 resolution switched back to connected22, but the backend connection remained to connected33. connected33 would have to drop the backend in this case to get things back on track.