12-13-2021, 07:53 PM
Sounds like bufferbloat at work. (Data quickly fills buffers then I/O slows down writing to the slow physical device.) However that shouldn't cause your restore to take over 5 hours.
I add "conv=fsync" to the dd command line to make sure data is flushed to the device before dd exits. When using a large buffer this can also help catch errors that might otherwise go unreported:
https://abbbi.github.io/dd/
As an aside, to save some disk space I pipe the image backup through gzip. If the pinephone is /dev/sdc this would be:
To restore:
I add "conv=fsync" to the dd command line to make sure data is flushed to the device before dd exits. When using a large buffer this can also help catch errors that might otherwise go unreported:
https://abbbi.github.io/dd/
As an aside, to save some disk space I pipe the image backup through gzip. If the pinephone is /dev/sdc this would be:
Code:
dd if=/dev/sdc bs=1M status=progress conv=fsync | gzip -c >pinephone.img.gz
To restore:
Code:
gunzip -c pinephone.img.gz | dd of=/dev/sdc bs=1M status=progress conv=fsync