On Mon, 2019-03-11 at 18:10 +0100, Matthias Klumpp wrote:
Am Mo., 11. März 2019 um 15:06 Uhr schrieb Jeremiah C. Foster jeremiah.foster@puri.sm:
[...] Using the above process filled up my disk. For some reason the pagemap file grew to 77 Gigs and on that partition it ran out of space.
-r-------- 1 root root 77G Mar 8 16:37 pagemap
I'll try on another (slightly larger) partition. How much disk space should one have for this? 77 Gigs seems like an awful lot of space for an image. All the source code for a image and yocto tooling in comparison comes to about 60 Gigs or so.
That's odd... The whole process defbitely works if you have much less then 4GB of disk space, and I think our current autobuilder has 30GB. What data was written to where, can you find that out?
The bulk of the data got written to that file above (pagemap), the rest of the file system was the same.
And which filesystem are you using?
findmnt in the container says ext4 as does the host.
# findmnt TARGET SOURCE FSTYPE OPTIONS / /dev/root[/var/lib/machines/pureos] ext4 rw,relatime,stripe=256 -host- / /dev/nvme0n1p3 ext4 rw,relatime,stripe=256
I assume you are running either PureOS or Debian Testing on the host system.
Yes;
# cat /etc/os-release PRETTY_NAME="PureOS" NAME="PureOS" ID=pureos HOME_URL="https://pureos.net/" SUPPORT_URL="https://puri.sm/faq/#faq-pureosandsoftware" BUG_REPORT_URL="https://tracker.pureos.net/"
You only need ~500MB to store a compressed root filesystem image, then some space to decompress it and the amount of space to build the image and space to store the resulting image. That should never fill up 77GB of space (and I built it with less then 4GB free on my local machine once).
Seems logical. Not sure what happned in my system. I'll run the commands again.
Regards,
Jeremiah