Error on API after upgrade to 1.7

[zoic22]

Hi,

I juste upgrade from 1.5.0 rc 1 to 1.7.0 and I have some trouble with api (which work fine in 1.5.0 rc1).
I only use js api with home software for home automation and since 1.7.0 some comande fail. My software is in php so I use curl to call zway, there is 4 retry (with 1 second between wich) before return command as failed. Each try have maximum 1 second to answer (I hope zway api can answer in less than one seconds). In 1.5.0 rc 1 I never got a failed and now with 1.7.0 I have 3 or 4 failed by day (most of it is timeout).

My command is like "turn on lamp", "turn off lamp". It should be noted that I also always a demon who asks every second if there are Zway events.

So there has there been changes that explain this or is this a regression?

[pofs]

Which requests exactly are failing for you?
There are some changes in the structure of the data holders, so some data paths may be different now. The rest of API is still the same.

[zoic22]

It's failling in this request for example :
http://192.168.0.12:8083/ZWaveAPI/Run/d ... 6].Set(99)

But not always only 3 or 4 time by day, the other time same request it's ok. It is as if the Zway was unavailable for ten seconds.

[pofs]

Is it actually "unavailable" (no response from server), or it returns HTTP error for that kind of request?
Unavailability might be caused by some network issues, not necessary related to z-way.

The only change to web server is enabling HTTP connection keep-alive feature, but it is unlikely to be a reason for connection failures.

[zoic22]

Here the exact error :
Operation timed out after 1001 milliseconds with 0 bytes received sur la commande http://192.168.0.12:8083/ZWaveAPI/Run/d ... 26].Set(0)

For me is not the network, my system work from 3 month correctly end juste after update it's fail.

Keep alive improves network response time but has a strong impact on the performance. So maybe it's can explain why I lost connection to razberry while few second. Can I disable keep alive to test ?

[pofs]

Passing "Connection: close" header will disable HTTP keep-alive.

[zoic22]

I already close connection at the end : curl_close($ch); . So maybe it's not that. But I have these error every day. Are they any test that I can perform on api to find the root of problem ?

[pofs]

Try looking into z-way server log file and match log entry with your failed request (by timestamp).
If there's a log entry, it means your request reaches server in time, but processing takes too long. If not, it is likely a networking issue.
Also try increasing timeout and see if the error is gone.

[zoic22]

Here the log from zway :
http://pastebin.com/zzytJhHS

The error occure at 27-05-2014 08:03:35 :
27-05-2014 08:03:35 error
Erreur sur Lumière : Echec de la requete http : http://192.168.0.12:8083/ZWaveAPI/Run/d ... 26].Set(0)
27-05-2014 08:03:35 error
Erreur sur Prise : Echec de la requete http : http://192.168.0.12:8083/ZWaveAPI/Run/d ... ].Set(255)
27-05-2014 08:03:35 error
Erreur sur Lumière : Echec de la requete http : http://192.168.0.12:8083/ZWaveAPI/Run/d ... 6].Set(99)

Time is sync between log and zway. What I can see it's the event take 30 sec to go on zwave :
[2014-05-27 08:04:01.563] Executing script: zway.devices[2].commandClasses[0x26].Set(0)
[2014-05-27 08:04:01.569] Executing script: zway.devices[11].commandClasses[0x25].Set(255)
[2014-05-27 08:04:01.566] Executing script: zway.devices[4].commandClasses[0x26].Set(99)

I revert back to 1.5.0 to test if it's definetly 1.7.0 version the root of api problem.

[alibab]

Hi,
I revert to 1.5 RC3 too.
The response with version 1.7 and openremote is very slower .
But work fine with 1.5....