Software Updates Timing Out

[watched730]

I have been having wireless problems and after reviewing and applying some of the troubleshooting steps and learning about doing a reset I have been trying to update to the latest "unstable" software to hopefully fix some of my issues.

Currently if i unplug and do a reset i will connect about 75 percent of the time and start a download. The download appears to stop at nearly the same spot every time. And eventually go back to the home screen, where it remains connected to access point and searching for server.

Other info:

• brewbit is 1 foot from wireless access point (latest airport extreme operating as bridge to comcast router)
• comcast router has static IP for brewbit on 192.168.1.X
• still on shipping software 1.0.0

Thanks in advance for ideas and help!

[ecotony]

I had this issue too. Thought it had updated, but was later disappointed to learn it had just reset.

I left it alone and tried later and it ultimately updated -- after 3 tries and keeping a closer eye on the MT.

Don't give up. Other option is to move the MT closer to your wireless router and give that a try.... seems the MT's wifi connection isn't the strongest and benefits from a close proximity to the router.

[watched730]

Thanks for the encouragement. I have reproduced the timeout at least 10 times with the wireless airport less than 2 feet away from the device. Sadly it never gets past 1/6 complete. It seems to download fine for about 90 seconds before progress halts.

After progress halts (but before timeout sends things back to home screen) I have checked other status on device using menus. If i navigate to connection section, it usually reports "connecting to server". If i go to the device info, it has the expected IP and network information.

[ecotony]

Sorry.

Maybe we have issues with some MTs or maybe the type of router?

[watched730]

Update:

I have reproed the timeout consistently 5 more times monitoring the download progress. It stopped at the same spot all 5 times. I also removed the airport (in bridge mode) and connected directly to the comcast wifi router and experienced the same behavior. I live in in SF so the wireless networks are pretty dense, but the consistent hang spot suggests to me its not the wifi stability. Still may be another local network issue. I will try to isolate that issue by taking the unit to a friends house and see what the behavior is.

If behavior is consistent at friends house I will be logging a "bug" thread as it seems like there is at least one other case of this behavior mentioned here. I have seen other threads talk about download timeouts for updates on server side on the "guinea pig" thread. @hebnern, It would be great to know if you are still tracking any server side issues.

FWIW the temperature controls (configured manually) seems to work great, so i am excited to get wifi working more consistently to take advantage of the full feature set. Consider me a guinea pig, happy to help troubleshoot if i can!

[hebnern]

Thanks for the info. The issue that I found that seemed to be causing other people to have download issues was the server was getting invalid messages that it was unable to decode. I assumed that this was due to the message stream getting out of sync. The "fix" that I added to address this was just disconnect the device and have it resync and try again. It is possible that the issue is not syncing, but that the device is actually sending a bad message or the server decoding is not handling a valid message properly. I will add more debugging to the server to dump out the message and see if I can figure out what might be going wrong.

[hebnern]

Can you try it again. I just added more debugging to the server to dump the message.

[cdburg]

I just tried to download the update. I've tried several times, and it seems to get about 10% of the way through, clicks and resets the unit. That's happened several times.

[cdburg]

I've tried about 10 times with no luck. Once, I got almost to 50% before the unit reset. That was as close as I got. Usually, it got no farther than 10% or so.

[hebnern]

Hmm. Ok, thanks for the data point. I am scouring the server logs for clues.

Make sure to bring your unit close to your router to make sure signal strength is not an issue.

[watched730]

Yep. Still seeing the same behavior as before too. It does seem pretty consistent on how far it gets though (i am within 2 feet from router).

Good luck digging through logs! Ill be up a bit if you need me to try again.

[cdburg]

If it helps, I finally got the update, on the first try, this morning.

[hebnern]

Awesome. I found and patched some issues with the server side message handling last night, so hopefully that was it. Thanks for the update.

[watched730]

Sounds like progress! An interesting data point. @cdburg, were you updating from 1.0.0?

Unfortunately I am still seeing the same behavior. A couple more data points. This morning i setup a static profile on the web and pushed it to model t. I had never tried this before. It worked first time no problems! I then monitored it to see how it did pushing data to the web (still within 2 feet of router). Seemed to do OK for a bit and then then it lost the connection. I timed the connection uptime and it was consistently 1 minute 45 seconds +/- a few seconds. As a second data point i timed how long it took for the the update download to stop, starting the clock on first connection. It was the same 1 minute 45 seconds.

SO.. maybe there is something with the 100 second or 2 minute mark in network or device polling?

I will take the device to a friends place tonight to see if i get the same behavior there.

[bmoney]

Thanks. If you continue to have issues we may just send you a new unit or controller board if you don't mind opening up the unit. We have read that not just our product but several other products that used TI's CC3000 WiFi module are having tons of issues similar to what we are experiencing.

[cdburg]

Yes, I was updating from 1.0.0.

[hebnern]

The CC3000's network connectivity tracking has been pretty unreliable in our testing. Because of this, we were seeing situations where the CC3000 was saying we are connected when we were really not. The only way to recover from this condition is to restart the WiFi module. So as a workaround, we are pinging google's public DNS server (8.8.8.8) to check for internet connectivity. If we are unable to reach it after 2 minutes, we restart the WiFi module. Perhaps this is the issue? Can you try pinging 8.8.8.8 from your computer?

[watched730]

Ran the ping.

It runs most of the time with a response of 35 ms. I did some long runs and occasionally (1 timeout per 150 or 200 requests) I see something like below where 1 - 2 requests will timeout, but eventually come in (summary shows no packet loss).

64 bytes from 8.8.8.8: icmp_seq=58 ttl=46 time=35.587 ms
64 bytes from 8.8.8.8: icmp_seq=59 ttl=46 time=35.674 ms
Request timeout for icmp_seq 60
Request timeout for icmp_seq 61
64 bytes from 8.8.8.8: icmp_seq=60 ttl=46 time=2059.877 ms
64 bytes from 8.8.8.8: icmp_seq=61 ttl=46 time=1058.863 ms
64 bytes from 8.8.8.8: icmp_seq=62 ttl=46 time=59.343 ms
64 bytes from 8.8.8.8: icmp_seq=63 ttl=46 time=35.303 ms
64 bytes from 8.8.8.8: icmp_seq=64 ttl=46 time=35.928 ms
64 bytes from 8.8.8.8: icmp_seq=65 ttl=46 time=35.303 ms

Also, been wondering this, but resetting firmware (with press 30 second hold) doesnt clear any settings. Is this expected behavoir?

[watched730]

Also ran it
- wired to comcast router. No timeouts
- from comcast wireless connection (rather than mac airport bridge) with similar results as above. It would time out .6 % of the time.

So in my dense wifi area it still completes 99 percent of the time. Dont know if you are looking for any drops over two minutes or checking a ping or two at 2 minute mark.

Let me know if you need me to check anything else.

[hebnern]

That should be fine. If we get even a single successful ping in 2 minutes, that will keep it connected. When we are not getting successful responses, we up the retry rate to 30 seconds to recover faster, but while things are up and running normally, we slow down to once a minute.