::open hangs sometimes

[tacoe]

Sometimes, calls to open - like wifly.open("api.cosm.com", 80) - hang and never return. I'll try using non-blocking and openComplete, but the blocking version could perhaps use a timeout?

[tacoe]

BTW, great library otherwise — right level of abstraction, too

[tacoe]

Bummer. Nonblocking has the same problem, so no way to circumvent.

Fragment below; Sometimes, last line never gets executed, effectively stopping everything without a way to intervene.

Serial.println("Opening server call (nonblocking)"); wifly.open("api.cosm.com", 80, false); Serial.println("Entering wait loop");

[harlequin-tech]

Hi tacoe,

that's a strange problem. The open() command should timeout after 10 seconds. I wonder if you're getting hit by the 2.32 firmware bug where the WiFly reboots? Depending on your shield, the WiFly reboot may reboot your Arduino. If it doesn't then I would expect the open() call to still fail with a timeout after 10 seconds.

I'll see if I can reproduce the behaviour you're seeing. If you have a small sketch that demonstrates the problem that would help. Are you using software serial for the WiFly interface?

Cheers.

[tacoe]

Hi, thanks for getting back!

Sketch (shortened to the essence) is here: https://gist.github.com/2900840

When ran, after anywhere from 5 minutes to 6 hours, the Arduino hangs. When I hook up a terminal, it looks like consistently, the hanging is in the ::open (which I deduce by putting logging commands immediately before and after it).

Since the terminal remains, it doesn't seem to be a reboot. That would actually be fine — some testing showed that WiFly reboots reboot the Arduino in the process, so no problems there. I'll verify my firmware version.

[tacoe]

Hmm. If I screen into the tty, the command prompt says 2.31 which I assume is the firmware version. Is the behavior described a known 2.31 issue? This is really not a question for you as it is for Roving Networks, I realize. Having looked at the WiFlyHQ code, my gut feeling is you can close this issue anway.

BTW, how do I enable/see the debug messages in your code?

[tacoe]

Gr. Still happening.

Serial.print("...connecting"); if(wifly.open(server,5000)){ Serial.println("...opened");

... hangs on .open; 'opened' never happens.

WiFly's D1 led is slow blinking during hanging. Can't find what that LED is -- it could mean either "Associated, no Internet" or "IP address OK" going on the WiFLy module manual, but that's a guess.

[harlequin-tech]

BTW, how do I enable/see the debug messages in your code?

Change line 37 in WiFlyHQ.cpp to "#define DEBUG". You may get timeouts due to too much debugging.

[harlequin-tech]

Could you try printing out the amount of free memory before each open attempt? I'm wondering if there is a low memory issue.

E.g.

Serial.print(F("Free memory: ")); Serial.println(wifly.getFreeMemory(),DEC);

[tacoe]

Will do and report back, thanks!

[harlequin-tech]

Are you still seeing this problem?

[harlequin-tech]

Very quiet. I'm going to assume the you tracked the problem down to an out-of-memory issue. Re-open the ticked if that's not the case.

[tacoe]

Fair enough, thanks for bearing with me.

[ghost]

Hi I also have this problem. It happen in Random like 24h-48h or 3 days after consecutive usage. I have firmware 2.36

Any ideas?

[harlequin-tech]

Hi vrgomes, have you checked your memory usage over 24 hours?

[ghost]

I didn't did that but the problem occurred somehow in random days. Maibe I have to make a log file with wifi working.

[harlequin-tech]

It might be a memory leak or out-of-memory problem. So checking the sketches memory use would be a good step.

[ghost]

Hi, I mesured memory just before it hanged and it was ok: freeRam() 3526

In the code I have: Serial.println("1"); if (wifly.open(site, 80)) { Serial.print("Connected to "); //more code...... } else { Serial.println("Failed to connect"); }

And in the console showed: 1

And it hanged there on wifly.open and stopped completely the arduino loop cycle.

I watched the code: boolean WiFly::open(const char *addr, int port, boolean block)

And it has: if (!getPrompt()) { debug.println(F("Failed to get prompt")); debug.println(F("WiFly has crashed and will reboot...")); while (1); /* wait for the reboot */ return false; }

That: while (1); /* wait for the reboot */

Isn't going to hang the code there for ever?

EDIT: I did some digging and the last debug print was: send_P: getPrompt "<2.36> " match: <2.36> readTimeout - timed out match: false Failed to get prompt WiFly has crashed and will reboot...

[rodri16]

I am with the same problem, I think there is a problem with while (1) boolean WiFly::open(const char *addr, int port, boolean block)

if (!getPrompt()) { debug.println(F("Failed to get prompt")); debug.println(F("WiFly has crashed and will reboot...")); while (1); /* wait for the reboot */ return false; }

[quique123]

Does upgrading the firmware solve the issue?