zram_tmp.service breaks my system about zram-init

[vaeth]

It works here without any problem.

Check whether "systemctl enable zram_tmp.service" actually created the correct symlink in /etc/systemd/system/local-fs-pre.target.wants

Since you use /etc/systemd/system/zram_tmp.service, I suppose that you modified it. Note that the default option -d1 requires that you set num_devices=2 (or larger) in /etc/modprobe.d/zram.conf

Perhaps journalctl _SYSTEMD_UNIT=zram_tmp.service will show you another cause of failure.

[morfikov]

Hmm

# cat /etc/modprobe.d/zram.conf
options zram num_devices=3

# systemctl mask tmp.mount
Created symlink from /etc/systemd/system/tmp.mount to /dev/null.

# systemctl enable zram_tmp.service
Created symlink from /etc/systemd/system/local-fs-pre.target.wants/zram_tmp.service to /etc/systemd/system/zram_tmp.service.

# ls -al /etc/systemd/system/local-fs-pre.target.wants/zram_tmp.service
lrwxrwxrwx 1 root root 36 2015-01-27 09:14:04 /etc/systemd/system/local-fs-pre.target.wants/zram_tmp.service -> /etc/systemd/system/zram_tmp.service

And it doesn't work. There's no log in journalctl _SYSTEMD_UNIT=zram_tmp.service .
It looks as if the service was invisible.

[morfikov]

Do I have to mask the tmp.mount ?/ I just did a small test, and I left tmp.mount unmasked, and it worked:

# mount | grep -i tmp
udev on /dev type devtmpfs (rw,relatime,size=10240k,nr_inodes=236995,mode=755)
tmpfs on /run type tmpfs (rw,nosuid,relatime,size=383388k,mode=755)
tmpfs on /dev/shm type tmpfs (rw,nosuid,nodev)
tmpfs on /run/lock type tmpfs (rw,nosuid,nodev,noexec,relatime,size=5120k)
tmpfs on /sys/fs/cgroup type tmpfs (ro,nosuid,nodev,noexec,mode=755)
/dev/zram1 on /tmp type ext4 (rw,discard)
tmpfs on /run/user/1000 type tmpfs (rw,nosuid,nodev,relatime,size=191696k,mode=700,uid=1000,gid=1000)

[vaeth]

Are you using the current systemd-218? IIRC some older version did not support local-fs-pre.target
No idea why it works for you if you don't mask: I would expect that tmp.mount overmounts your /tmp with a tmpfs - after all, this is the purpose of tmp.mount.

[morfikov]

I was using 215, but I thought maybe it's not enough, so I upgraded it to 218.

[morfikov]

I know where the problem was, and I have no idea how this could happened, but somehow I left /tmp entry in the /etc/fstab file uncommented . When I commented that out and masked tmp.service, and then I enabled zram_tmp.service, everything was just fine after reboot.