Freezing Rock 5B with Rsync transfer

Appreciate posting this. I had a similar problem using FTP protocol where it would simply shut down network after a few minutes of transfer.

what’s the iamge you used, is also the armbian ?

we will check this issue

Issue on debian, ubuntu server and armbian. Was a while ago so I should retest latest images. Appriciate taking a look…

First, for the god sake, use spoilers for logs

Second, retest it with Radxa images

Third, does it happens if you do microSD/flash drive transfer?

Fourth, does it happens if you rsync from nvme to nvme?

Does it happens if you dd file from /dev/urandom to nvme? Something like
sudo dd if=/dev/urandom of=/test.img bs=1m count=65536

Fifth, does it happens with cp?

Just out of curiosity.

ouch, never mind, 1.5A is the input. Ignore.

But if this 65W is like the one i tested 33W, it could negotiate 20v, 1.35A or something…

See here: USB-C PD power charger behavior

Try images with kernel 6.5.y at the bottom of the page. You need to boot from SD card and there is no video out. Use serial console or SSH access …

Allright thanks :slight_smile:

Rsync transferring fix was :frowning:

Limiting RAM frequency improves stability for me and stops the random hangs/reboots:
echo 1560000000 | sudo tee /sys/ class / devfreq / dmc / max_freq

1 Like

Known problem on some boards: https://github.com/ThomasKaiser/sbc-bench/issues/55

But that this happens with rsync is somewhat strange? Can you test through sbc-bench -s walking through the DMC frequencies from bottom to top? At which RAM clock is your board freezing?

Will this also be remembered if rebooting the device? do we need to set governor userspace and how?

I also had to reduce cpu frequency sigh. Was this hardware ready when it launched?? (yes i bought mine as some of the early models) I mean happy it seems to be working so far but kinda sad most of my HW must be run outside of specs to be able to function.

1 Like

Hmm I said that too early even lowering to 2ghz it crashed when using WinSCP same time I was trying to browse a plex server.

update: with latest debian image it is crashing still but it seems its able to reconnect every time some seconds later at least so not a full network shutdown anymore. This is using the official raddax power supply so not sure if that has helped too.

Also if it is helpfull I transfered a few books I have bought from humblebundle on a another rock pi 5b using calibre web and this did not fail a single transfer. I am wondering if it is related to larger files and that it is then related to some memory problem? But why would it then shut down the network ?

hey there,
this problem still persists !
I’m using a 65Watt power supply and also added a heatsink around the nvme …
network ping is still answering, but the rock5b is not accessable anymore.
this is a hard problem, I bought this to replace my raspi 4.
but with this performance issues … no way .

1 Like

Have you tired this method

"
Limiting RAM frequency improves stability for me and stops the random hangs/reboots:
echo 1560000000 | sudo tee /sys/ class / devfreq / dmc / max_freq
"

Known problem on some boards: https://github.com/ThomasKaiser/sbc-bench/issues/55

Today I was down to 528MHz, the upload with winscp survives a little longer, but after about an hour the rock5b died again….
could someone try it on their machine to reproduce it and maybe find a solution?

just to add, I have a rock5b with 32GB ram and a samsung 980 1TB NVMe. also tried different OS with different kernel versions.
looks like, if I limit the upload to 1MBit/s it works ….
But the rock5b has what, 2.5GBit ethernet …

It might be the SSD. 980s are quite power hungry and Rock 5b was reported to hang when using them. Try using a less top of the line SSD and it will never hang any more.

Hi There, you may want to try this: Radxa Rock 5b Hangs multiple times a week

BTW: Depends on which win version you are using but in my win 11 i can use scp from terminal without needing putty / winscp.
Best

Now after testing and try and and …
I monitored the system all the time an now I found this.
if the RAM Cache + Buffer metrics reaches 10.3 GiB it freezes. It’s reproducable. Now I need an idea how to overcome this border

If I cleanup the buffer / cache if it reaches 10GB … the rock5b will survive … he’s not freezing …
so maybe my memory is faulty or something else.
I will contact the supplier and ask for a new one …

Maybe others with this issue can try to reproduce it ?!
and give a feedback