mesh serial DFU lost connection

[Mike]

"Failed to upgrade target. Error is: Lost connection in the middle of the transfer"

I am doing a mesh serial DFU with DFU quick start guide.

I have gotten all the way to step 10, performing the DFU. The DFU process keeps cutting out partway. Sometimes it gets to 2%, sometimes it doesn't get to 1%. I have checked on-chip debug logs (with DFU messages filtered in) and I see nothing helpful there. I need a lead to explore. Any suggestions?

A message tx from pc is not responded to and retransmitted, at which point error responses start. I have tried intervals from 20ms to 2000ms. 

B:\Devel\_LocalCopy\PCS_SERIAL 1.07 DFU\tools\dfu>nrfutil --verbose dfu serial -pkg dfu.zip -p COM9 -b 115200 -fc -i 20 --mesh
Upgrading target on COM9 with DFU package B:\Devel\_LocalCopy\PCS_SERIAL 1.07 DFU\tools\dfu\dfu.zip. Flow control is enabled.
[------------------------------------] 0%Flushing com-port...
Opened com-port
Starting DFU upgrade of type 4, SoftDevice size: 0, bootloader size: 0, application size: 140868
Sending DFU start packet, afterwards we wait for the flash on target to be initialized before continuing.
PC -> target: 0502aabbccdd
target -> PC: 0582aabbccdd
Got echo response
Sending DFU init packet
PC -> target: 1378fdff040fb48dd2f153435000010007000000
target -> PC: 16a6045343500001000700000053435000010001000000
target -> PC: 03847800
PC -> target: 1378fdff040fb48dd2f153435000010007000000
target -> PC: 03847800
PC -> target: 1478fcff0000b48dd2f1ffffffff9189000040000c
target -> PC: 0da2010453435000010007000000
target -> PC: 03847800
Sending firmware file
[------------------------------------] 1% 00:11:56PC -> target: 1978fcff0100b48dd2f100f00020356202006562020067620200
target -> PC: 03847800
PC -> target: 1978fcff0200b48dd2f1696202006b6202006d62020000000000
target -> PC: 03847800
PC -> target: 1978fcff0300b48dd2f10000000000000000000000006f620200
target -> PC: 03847800
PC -> target: 1978fcff0400b48dd2f171620200000000007362020075620200
target -> PC: 03847800
PC -> target: 1978fcff0500b48dd2f1796202007b620200a93c04007f620200
target -> PC: 03847800
PC -> target: 1978fcff0600b48dd2f181620200836202006186020087620200
target -> PC: 03847800
PC -> target: 1978fcff0700b48dd2f1896202008b620200618a03008f620200
target -> PC: 03847800
PC -> target: 1978fcff0800b48dd2f191620200936202009562020097620200
target -> PC: 03847800
PC -> target: 1978fcff0900b48dd2f199620200f1c0030029ce02009f620200
target -> PC: 03847800
PC -> target: 1978fcff0a00b48dd2f119c10300a362020091190400a7620200
target -> PC: 03847800
PC -> target: 1978fcff0b00b48dd2f1a9620200ab620200ad620200af620200
target -> PC: 03847800
PC -> target: 1978fcff0c00b48dd2f1b1620200b36202007762020077620200
target -> PC: 03847800
PC -> target: 1978fcff0d00b48dd2f1b5620200b7620200b9620200bb620200
target -> PC: 03847800
PC -> target: 1978fcff0e00b48dd2f1bd620200bf620200c162020035493648
target -> PC: 03847800
PC -> target: 1978fcff0f00b48dd2f10a1a02d0072291438d46344934480a1a
target -> PC: 03847800
PC -> target: 1978fcff1000b48dd2f106d00722914381f30988022282f31488
target -> PC: 03847800
PC -> target: 1978fcff1100b48dd2f130483149314a00f043f831483149324a
PC -> target: 1978fcff1100b48dd2f130483149314a00f043f831483149324a
target -> PC: 03847800
target -> PC: 03840084
PC -> target: 1978fcff1200b48dd2f100f03ef831483249324a00f039f83248
PC -> target: 1978fcff1200b48dd2f100f03ef831483249324a00f039f83248
PC -> target: 1978fcff1200b48dd2f100f03ef831483249324a00f039f83248
PC -> target: 1978fcff1200b48dd2f100f03ef831483249324a00f039f83248
PC -> target: 1978fcff1200b48dd2f100f03ef831483249324a00f039f83248


Failed to upgrade target. Error is: Lost connection in the middle of the transfer

[Bjørn Kvaale]

Have you tried running the nrfutil command without the interval option (i.e. -i 20)? Does that also fail before the DFU completes?

From experience, usually there is one small issue somewhere in the DFU quick start guide steps. If you have not created a script for this process, I would highly recommend doing this first. If that doesn't work, let me know & I'll take a closer look.

Are you using a nRF52 DK as the serial device that communicates with the mesh network?

[Mike]

Good suggestions, but I already did both. I had many small issues with the DFU quick start guide, and scripts fixed some of them. I have also tried intervals of 500, 1000, and 2000.

...

I figured it out! The flow control in my device is turned off and a byte is lost in transmission. Then the serial stream gets misaligned. I had to modify the nrfutil so that when it doesn't get a response, it flushes the target's buffer with 0x00 until it gets a response. It looks like this now (a snip from the middle of a TRX):

target -> PC: 03847800
PC -> target: 1978fcff11001adb8bda30483149314a00f043f831483149324a
target -> PC: 03847800
PC -> target: 1978fcff12001adb8bda00f03ef831483249324a00f039f83248
target -> PC: 03847800
PC -> target: 1978fcff13001adb8bda3249334a00f034f832483349334a00f0
target -> PC: 03847800
PC -> target: 1978fcff14001adb8bda2ff833483349344a00f02af833483449
PC -> target: 0x00,  #0
PC -> target: 0x00,  #1
PC -> target: 0x00,  #2
PC -> target: 0x00,  #3
PC -> target: 0x00,  #4
target -> PC: 03847800
target -> PC: 03840084
PC -> target: 1978fcff15001adb8bda344a00f025f834483449002200f02bf8
target -> PC: 03847800
PC -> target: 1978fcff16001adb8bda33483449002200f026f833483349091a
target -> PC: 03847800
PC -> target: 1978fcff17001adb8bda082903db00220260043001601f482049
target -> PC: 03847800
PC -> target: 1978fcff18001adb8bda884205d00268043003b4904703bcf7e7
target -> PC: 03847800
PC -> target: 1978fcff19001adb8bda00208646ec4600200021294a9047fee7
target -> PC: 03847800
PC -> target: 1978fcff1a001adb8bda884207d0521a05d0037801300b700131
target -> PC: 03847800
PC -> target: 1978fcff1b001adb8bda013af9d17047884202d002700130fae7
PC -> target: 0x00,  #0
PC -> target: 0x00,  #1
PC -> target: 0x00,  #2
PC -> target: 0x00,  #3
PC -> target: 0x00,  #4
PC -> target: 0x00,  #5
PC -> target: 0x00,  #6
PC -> target: 0x00,  #7
PC -> target: 0x00,  #8
PC -> target: 0x00,  #9
PC -> target: 0x00,  #10
PC -> target: 0x00,  #11
PC -> target: 0x00,  #12
PC -> target: 0x00,  #13
PC -> target: 0x00,  #14
PC -> target: 0x00,  #15
PC -> target: 0x00,  #16
PC -> target: 0x00,  #17
PC -> target: 0x00,  #18
PC -> target: 0x00,  #19
target -> PC: 03847887
PC -> target: 1978fcff1b001adb8bda013af9d17047884202d002700130fae7
target -> PC: 03847800
PC -> target: 1978fcff1c001adb8bda7047000000f0002000e8002000f00020
target -> PC: 03847800
PC -> target: 1978fcff1d001adb8bda00f00020c8840400c832002044340020
target -> PC: 03847800

It took about 12 minutes.

[Bjørn Kvaale]

Glad to hear that you figured it out! The Mesh DFU can take a while if you are updating multiple mesh nodes at the same time, as the throughput is not that high.

[Mike]

Correction: it takes around 35 minutes to do a single upgrade, and not all devices succeed.

It will be faster to just serial upgrade all devices at a faster interval. How can I turn off the mesh portion of the upgrade?

[Bjørn Kvaale]

Very sorry for the late response. You can use nrf_mesh_dfu_jump_to_bootloader() to jump to the bootloader & run a bootloader mode DFU. The device will then not participate in the mesh network again until the DFU is successful (see link). More documentation is available here.