Hello @avrobot
I was able to have Mender built with Yocto for Colibri iMX6ULL following the work done on the pull request:
But you have to consider how much space on flash you need. In my case, after adding all packages and applications for my product, I only had 6MB left.
That’s because Mender uses an A/B partition solution.
Hi @qba,
Please, next time create a new post so we can address your support directly, and then, you can also mention this ongoing post, for example.
Regarding your issue:
It is possible to update system from any medium: www, mmc, usb stick etc.
That depends on SWUpdate.
Having a look at SWUpdate Git, it tells a lot about this:
Multiple interfaces (local and OTA) for getting software
So, judging by SWUpdate, yes, it’s possible.
But if you face any issues, then I suggest you open an issue request on SWUpdate Git.
Best regards,
André Curvello
Hi @avrobot,
Can you confirm if these messages appear as soon as you reboot the board again?
I saw some references to this issue here and here.
From the second link:
It look likes the Ubi error only appears in the first time the board boots after reflash (run update && run setupdate). If after the first time I reboot the board and try to git clone on the second time linux boots it works.
Of course, this is a surprise, and we expected it do not appear. But judging by an ubifs developer feedback on the 1st link, it seems to be inoffensive.
I saw that your original post was targeting BSP 2.8. I hope you don’t be mad about my suggestion, but I really suggest you move forward to BSP 3.0.4, because is our current LTS with a lot of improvements, bug fixes, etc…
The BSP 2.8b7 is only for production devices receiving security updates.
For newer products, BSP 3.0.4 is strongly recommended.
Best regards,
André Curvello.
Hi,
Thanks for the links!.. Slightly encouraging that one of the links mentioned its not a concern…
These messages only appear once after mender reboots when it finishes re-flashing an update. Actually we are using BSP 3.0.4b.
…but there will always be some worry unless it can be solved (or fully understood). What if in a year or two the NAND can’t be updated and we have bricked units in the field - this kind of recall can bring down a company.
The log is not clear what root cause could be.
Hi @avrobot,
As this topic was originally created by @HansKim72, can you create a new one, providing as much as possible details for your scenario, please?
Best regards,
André Curvello
