Dear all, there seems to be a regression in FBSD-10 w.r.t. ZFS+GELI setups again. For a reference to the first time this happened: http://lists.freebsd.org/pipermail/freebsd-fs/2013-July/017899.html . After that time, I added (as recommended) the previously default lines on the zpool caches to /boot/loader.conf. zpool_cache_load="YES" zpool_cache_type="/boot/zfs/zpool.cache" zpool_cache_name="/boot/zfs/zpool.cache" The setup is as follows: one drive, two ZFS pools, one for /boot (bootdir) and one on a GEOM_ELI layer for the rest of / (zroot). This setup is probably very close to what e.g. pjd_at_ runs according to above reference (so it is not quite the oddball config it was made to sound like in above thread). /boot then is a softlink to the other pool (bootdir/boot) After upgrading to r255479 from a maximum two weeks old CURRENT the following happens: * kernel loads and boots fine (including modules specified in loader.conf) * DRM load fails, iwn firmware load fails * VESA based Xorg works * /boot is a broken symlink as bootdir vanished, zpool list only shows the GELI based pool Same happens in single user mode. I can zpool import the missing pool (at least something) after that. Could somebody help me to get my bootdir back permanently and at boot time? That would be great since as of now, I am stuck with VESA mode graphics and no WLAN. Also, if this continues to break at random times: can I consider this setup officially unsupported and advised against? If so, it would be great if there was an officially recommended setup for this rather standard problem (notebook: one drive, all ZFS, encryption wanted)? Thanks a lot JohannesReceived on Tue Sep 17 2013 - 17:17:49 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:41 UTC