@Malte after a lot of faffing about, here’s my hypothesis… if the LTE network takes a bit before it starts letting data through (i guess it happens with some providers), minion tries to contact master and fails. And if it fails, it keeps on failing for some reason (same BadStatusLine error) forever.
Solution - ping master and wait for it to respond (indicating LTE network is finally letting data through) before bringing minion up.
Additionally, the power timers don’t get suspended during updates, causing issues on EVs or when running on a bench. Device should be always on while the update is running.
All of this is just guesswork, though, so i’m probably completely off.
I’ve resolved all my issues except the original one - i need to reboot the device multiple times until i luck out and have the cloud working.
Still no go for SMS wakeup after last update. Modem power save is off, connection to cloud and wifi OK. Is there a way to rollback to an older update without doing a reflash?
Upon browsing through the Autopi github repo, turns out i was wrong with the assumption that it doesn’t kill the power management during update state. It does (it deletes all queued up state changes before an update), but for some reason, that didn’t work in my case and it went through a few retries with sleep in between before it finally updated.
Also found out it brings up minion before the network is up (on purpose, for some reason). So if network not being up at the right time were the issue, it probably would’ve been an issue everywhere / for everyone.
So… still no idea why it’s having so much trouble talking to the cloud.
To make things more puzzling, whenever it happens, i can ping hub01.autopi.io just fine from the dongle. It’s just that i keep getting the same “Connection aborted” from cloud.status over and over and over again.
First boot (since Apr 14th) - connection to cloud working.
Let it go to sleep, woke it up - “Connection aborted”.
Another sleep, another wakeup - “Connection aborted”.
Hibernate, then wakeup - “Connection aborted”.
Complete power down, then on - “Connection aborted”.
4G, Wifi, hotspot, everything working fine except for the connection to cloud/master.
I’ve copied over the old minion credentials, so no “re-accepting in the AutoPi Cloud” was necessary.
Besides, as i’ve mentioned multiple times now, it sometimes connects to the cloud.
In about 1/10 boots.
On further research, the BadStatusLine exception comes from the Python request package.
It typically indicates it has received a non-parsable response from the server, or an empty one.
Could you check your access logs to verify what’s happening? I can provide timestamps if needed.
@Peter and @Malte , I must underline to you that is 3 weeks of my autopi died, I send on this blog info , I send logos on email , but I no have reply.
It’s not a good support this, therefore please, you can not sell products and not have a RMA procedure, also how can advice your products if support is leaked as I see.
I’m little bit astonished of this, if I can’t resolve I must use PayPal as third party but is force brute procedure and I not want use, but how can I do ?
I’m not completely sure what the problem is with your device. I’ll follow up with support to check on your case, but from what I can see the last reply was received from you just before the weekend. Unfortunately we dont have 24 hour support, or support out of business hours, so you need to expect a couple of days for use to investigate and reply to your case.
I’m glad that you got it working, and thank you for reporting with details, we will look into if the device needs to report more feedback to prevent other people from having the same issue.
Can you confirm that the issue was that the sim card required a pin number, and that the error was not shown in the local configuration tool?
