You are not logged in.
Starting Import ZFS pools by cache file...
[ 4.966034] VERIFY3(0 == zap_lookup(ddt->ddt_os, ddt->ddt_spa->spa_ddt_stat_object, name, sizeof (uint64_t), sizeof (ddt_histogram_t) / sizeof (uint64_t), &hht->ddt_histogram[type][class])) failed (0 == 6)
[ 4.966100] PANIC at ddt.c:124:ddt_object_load()
[*** ] A start job is running for Import ZFS pools by cache (Xmin Ys / no limit)
And then occasionally I see
[ 240.576219] Tainted: P O 3.19.2-1-ARCH #1
Anyone else experiencing the same?
Offline
Is it possible to import the pool while booted into the archiso? See https://wiki.archlinux.org/index.php/Un … po-archiso
Offline
Hi,
after updating to kernel 3.19.2 today I also cannot boot. Importing my ZFS pools from cache file seems to fail, but for another reason:
Starting Import ZFS pools by cache file...
[ 3.155488] VERIFY3(range_tree_space(rt) == space) failed (2096150 == 8585830400)
[ 3.155509] PANIC at space_map.c:127:space_map_load()
[*** ] A start job is running for Import ZFS pools by cache (Xmin Ys / no limit)
I don't know if our problems are related. Did you make any progress?
demizer: I was able to boot into a freshly built archiso (also with kernel 3.19.2 and newest zfs- and spl-packages) and import the pool by using the -f flag without problems.
Last edited by robm (2015-03-20 17:51:33)
Offline
I, too, was able to import the pool using archiso, but I did not need to pass the "-f" flag. I then disabled "zfs.target", which allowed the system to boot. Then I re-imported the pool, which worked. So I tried re-enabling zfs.target, rebooted, and now it's working!
Offline
Thanks!
I did the same and it worked... kind of. The first three reboots it failed (but did not stop the system from booting) producing:
zpool[426]: cannot import 'data': one or more devices is currently unavailable
systemd[1]: zfs-import-cache.service: main process exited, code=exited, status=1/FAILURE
The second boot also resulted in a kernel panic, but as far as I can tell unrelated to zfs.
After reboot one and three imported the pool manually.
From the fourth reboot on loading from cache file always succeeded. However, I it takes faily long (~8 seconds) and even shows
[*** ] A start job is running for Import ZFS pools by cache (Xmin Ys / no limit)
briefly. Altough I might only notice that because the recent updates sped up oder parts of the boot process. Did you observe a slowdown during boot time, too, kinghajj?
Last edited by robm (2015-03-22 01:21:05)
Offline
Okay I think the first failed zfs-import-cache were a result of accidentally reimporting the pool not by id, but by /dev/sd*. During first reboot I still had the live-usb-stick plugged in and removed it later which probably messed up the order of the drives regarding /dev/sd*.
Last edited by robm (2015-03-22 14:15:59)
Offline
Hey all,
I am having the same issues you guys are having with ZFS after updating to 3.19.2.
I am not a Linux guru by any means.
So may I humbly ask for a bit more information on your steps in solving the problem? I can deduce that you made a fresh archiso with zfs on it but I am getting no where in creating my own.
My arch linux box is my only linux box. I can get a live install of something if i need to.
Offline
I was able to chroot into my system and disable zfs.target
Now I have to somehow reenable it. Everytime I've tried to start or enable the zfs service it fails with the same error message robm received:
PANIC at space_map.c:127:space_map_load()
If this is related to the zfs cache, could I just delete .cache file?
Not sure where to go from here. I've been struggling to build an archiso with zfs since I can get the local signing for pgp keys to work right from an iso environment.
Offline
So I managed to get a zfs enabled iso environment going.
What is really strange is that I had my devices in my pool identified with /dev/disk/by-id.
Now they are by device. And perhaps that is why it was messing up the import on boot?
Offline
Fixed:
Use archiso with ZFS to force import the pool (probably should use import -d /dev/disk/by-id if you want it by id)
Export the pool
Reboot into the actual system
Delete etc/zfs/zpool.cache (I could because I only had one zfs pool in my system)
Import the pool by id again!
Offline