clusterhat v2.5 not working

[Marco Picard]

Hi, i start a thread to find a solution to resolve my issue with the clusterhat.

My setup is :

1 CM4 => 4G RAM => Lite => no wireless

1 CM4-IO Board

1 clusterhat v2.5

4 pi zero v1.3

controller image: bullseye 11 cbridge arm64 lite

pi zero images: bullseye 11 armhf lite

When the assembly is powered,  the blue led alert light on and doesn't shut off even at the end of the controller boot sequence. I know that is it not normal, i looked around on the net and i saw that the clusterhat do not initialize because of that. I verified the good connection of the clusterhat on controller and it is not the problem. 

[Chris Burton]

Hi, 

When the assembly is powered,  the blue led alert light on and doesn't shut off even at the end of the controller boot sequence. I know that is it not normal, i looked around on the net and i saw that the clusterhat do not initialize because of that. I verified the good connection of the clusterhat on controller and it is not the problem. 

Does "ls /proc/device-tree/hat/product" find a file or not on the controller?

Does "i2cdetect -a -y 1" show a device at 0x20?

What happens when you run "sudo clusterhat init"?

Chris.

[Marco Picard]

[Marco Picard]

[Chris Burton]

Hi,

Thanks - can you try adding  "dtparam=i2c_vc=on" to the end of /boot/config.txt, reboot and then run "i2cdetect -a -y 0" - does it show a device on 0x50.

If not can you try uncommenting the "CLUSTERHAT_FORCE=" line in /etc/default/clusterctrl and changing it to "CLUSTERHAT_FORCE=2.5", then can you run "clusterctrl status" or "sudo clusterctrl init" ?

Chris.

[Marco Picard]

good news, the process has initiate the clusterctrl.

but, when i launch a node "clusterctrl on p1" i waited for a ten minutes, and i did not get trace of node on the dhcp lease ipv4 address list.

i tried "lsusb -t" and i get this: 

[Marco Picard]

this is a topo of the network :

[Chris Burton]

Hi,

 Did "i2cdetect -a -y 0" show a device on 0x50 ? Adding the line to the /etc/defaults/clusterhat file doesn't fix the problem of the ClusterHAT not being detected it just works around it.

i tried "lsusb -t" and i get this: 

I assume the last "Driver=hub/4p, 480M" line is the USB hub on the Cluster HAT - if you remove the USB cable temporarily does it go away and come back when you plug it back in again? If not then check the USB cable going between CM4IO and Cluster HAT.

good news, the process has initiate the clusterctrl.

but, when i launch a node "clusterctrl on p1" i waited for a ten minutes, and i did not get trace of node on the dhcp lease ipv4 address list.

I'd advise running through the troubleshooting guide now ( https://8086.support/content/23/115/en/troubleshooting-missing-pi-zeros-in-a-clusterhat.html ) - it might be that you haven't pushed the Pi Zeros down far enough onto the uUSB connectors or there's a problem with the SD cards - the troubleshooting guide will tell you which.

Chris.

 

[Marco Picard]

the first cmdline is with usb cable to clusterhat

the second, without.

then lsusb -t show the presence of the clusterhat

I ll go right now to the guide to get more details.

[Marco Picard]

the blue alert led comeback, i tried to reimage my controller SDCard adding parameters in config.txt, without results, i wrote to pishop support team, they offer to support on RPI, then i will wait for what they can do for the clusterhat blue led always on. I ll take you informed of the progress. thanks Chris, for your time and patience. 

Marco

[Marco Picard]

hi Chris,

today, on Monday, pishop return a message to me and they change the support offer to limited replacement of a defective clusterhat. they told to ask to 8086 consultant to diagnose and if the clusterhat is defect, then they will replace it. would you please help me to analyse the clusterhat to know if it is defective

the serial number of the clusterhat is 1020006083 as mentioned on the sticker barcode

i can configure my router to let you ssh to the cm4io board if its needed

Marco

[Chris Burton]

Hi, 

today, on Monday, pishop return a message to me and they change the support offer to limited replacement of a defective clusterhat. they told to ask to 8086 consultant to diagnose and if the clusterhat is defect, then they will replace it. would you please help me to analyse the clusterhat to know if it is defective

As discussed on Discord the Cluster HAT is faulty. If you'd like me to send you a replacement direct please email your name/address/serialno details to sup...@8086.net.uk.

Chris.

[Marco Picard]

today, on friday, i mounted the replacement clusterhat newly received from Pishop. It is fully operational. Thank you very much.

Marco