On 26/07/21 5:42 pm, Geert Stappers wrote:
On Sun, Jul 25, 2021 at 03:43:18AM +1200, Richard Hector wrote:Hi all, Sorry for the minimal detail - hopefully someone else will understand this much better than I do, and be able to fill in if required. Failing that, I might be able to do better later. Just spent many more hours on this than anticipated, and need sleep. I installed bullseye on my HP ProBook 430 G3 (attempting to encrypt everything including /boot, but backed out of that - I don't think it was relevant in the end) - using UEFI. Grub failed to install. I looked at the logs on VT 3, and saw stuff about No space left on device. df showed my devices all seemed fine. It turned out that my NVRAM was full of dump files. I don't know what they're for and why they had accumulated, but a page on the Arch wiki suggested deleting them, which allowed grub to install. Is it worth putting comments about that in the release notes?The way to go is having a closer look, finding out what is caused it, solving the problem.
Thank Geert,Yes ... that might solve my particular problem. But I'm not sure where to start researching, especially since I deleted the dump files, and they don't appear to have come back, which means I'm unlikely to be able to reproduce the problem either.
The only references I can find to these dump files (eg https://wiki.archlinux.org/title/Unified_Extensible_Firmware_Interface) relate to deleting them if required, rather than explaining where they came from.
Partly I was hoping that somebody here understands UEFI and efivars better than I do (it wouldn't be hard).
Also, unrelated to my specific issues with my specific laptop, I wondered if it was worth detecting this condition, perhaps from grub-install, and displaying a message that might help the next person make progress. After all, a message that 'No space left on device' when the 'device' in question doesn't even show up in the output of df (why?), is of limited value.
Should I file a bug against grub-install (or grub2-common?)?