nixpkgs/nixos/modules/tasks/filesystems/zfs.nix

344 lines
11 KiB
Nix
Raw Normal View History

nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
{ config, lib, pkgs, utils, ... }:
#
# todo:
# - crontab for scrubs, etc
# - zfs tunables
2012-12-04 18:17:54 +00:00
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
with utils;
with lib;
2012-12-04 18:17:54 +00:00
let
2012-12-06 23:12:10 +00:00
cfgSpl = config.boot.spl;
cfgZfs = config.boot.zfs;
cfgSnapshots = config.services.zfs.autoSnapshot;
2012-12-04 18:17:54 +00:00
inInitrd = any (fs: fs == "zfs") config.boot.initrd.supportedFilesystems;
inSystem = any (fs: fs == "zfs") config.boot.supportedFilesystems;
enableAutoSnapshots = cfgSnapshots.enable;
enableZfs = inInitrd || inSystem || enableAutoSnapshots;
2012-12-04 18:17:54 +00:00
kernel = config.boot.kernelPackages;
splKernelPkg = if cfgZfs.useGit then kernel.spl_git else kernel.spl;
zfsKernelPkg = if cfgZfs.useGit then kernel.zfs_git else kernel.zfs;
zfsUserPkg = if cfgZfs.useGit then pkgs.zfs_git else pkgs.zfs;
autosnapPkg = pkgs.zfstools.override {
zfs = zfsUserPkg;
};
zfsAutoSnap = "${autosnapPkg}/bin/zfs-auto-snapshot";
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
datasetToPool = x: elemAt (splitString "/" x) 0;
fsToPool = fs: datasetToPool fs.device;
zfsFilesystems = filter (x: x.fsType == "zfs") (attrValues config.fileSystems);
isRoot = fs: fs.neededForBoot || elem fs.mountPoint [ "/" "/nix" "/nix/store" "/var" "/var/log" "/var/lib" "/etc" ];
allPools = unique ((map fsToPool zfsFilesystems) ++ cfgZfs.extraPools);
rootPools = unique (map fsToPool (filter isRoot zfsFilesystems));
dataPools = unique (filter (pool: !(elem pool rootPools)) allPools);
2012-12-04 18:17:54 +00:00
in
{
###### interface
options = {
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
boot.zfs = {
useGit = mkOption {
type = types.bool;
default = false;
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
example = true;
description = ''
Use the git version of the SPL and ZFS packages.
Note that these are unreleased versions, with less testing, and therefore
may be more unstable.
'';
};
extraPools = mkOption {
type = types.listOf types.str;
default = [];
example = [ "tank" "data" ];
description = ''
Name or GUID of extra ZFS pools that you wish to import during boot.
Usually this is not necessary. Instead, you should set the mountpoint property
of ZFS filesystems to <literal>legacy</literal> and add the ZFS filesystems to
NixOS's <option>fileSystems</option> option, which makes NixOS automatically
import the associated pool.
However, in some cases (e.g. if you have many filesystems) it may be preferable
to exclusively use ZFS commands to manage filesystems. If so, since NixOS/systemd
will not be managing those filesystems, you will need to specify the ZFS pool here
so that NixOS automatically imports it on every boot.
'';
};
forceImportRoot = mkOption {
type = types.bool;
default = true;
example = false;
description = ''
Forcibly import the ZFS root pool(s) during early boot.
This is enabled by default for backwards compatibility purposes, but it is highly
recommended to disable this option, as it bypasses some of the safeguards ZFS uses
to protect your ZFS pools.
If you set this option to <literal>false</literal> and NixOS subsequently fails to
boot because it cannot import the root pool, you should boot with the
<literal>zfs_force=1</literal> option as a kernel parameter (e.g. by manually
editing the kernel params in grub during boot). You should only need to do this
once.
'';
};
forceImportAll = mkOption {
type = types.bool;
default = true;
example = false;
description = ''
Forcibly import all ZFS pool(s).
This is enabled by default for backwards compatibility purposes, but it is highly
recommended to disable this option, as it bypasses some of the safeguards ZFS uses
to protect your ZFS pools.
If you set this option to <literal>false</literal> and NixOS subsequently fails to
import your non-root ZFS pool(s), you should manually import each pool with
"zpool import -f &lt;pool-name&gt;", and then reboot. You should only need to do
this once.
'';
};
};
services.zfs.autoSnapshot = {
enable = mkOption {
default = false;
type = types.bool;
description = ''
Enable the (OpenSolaris-compatible) ZFS auto-snapshotting service.
Note that you must set the <literal>com.sun:auto-snapshot</literal>
property to <literal>true</literal> on all datasets which you wish
to auto-snapshot.
2012-12-04 18:17:54 +00:00
You can override a child dataset to use, or not use auto-snapshotting
by setting its flag with the given interval:
<literal>zfs set com.sun:auto-snapshot:weekly=false DATASET</literal>
'';
};
2012-12-04 18:17:54 +00:00
frequent = mkOption {
default = 4;
type = types.int;
description = ''
Number of frequent (15-minute) auto-snapshots that you wish to keep.
'';
};
hourly = mkOption {
default = 24;
type = types.int;
description = ''
Number of hourly auto-snapshots that you wish to keep.
'';
};
daily = mkOption {
default = 7;
type = types.int;
description = ''
Number of daily auto-snapshots that you wish to keep.
'';
};
2012-12-04 18:17:54 +00:00
weekly = mkOption {
default = 4;
type = types.int;
description = ''
Number of weekly auto-snapshots that you wish to keep.
'';
};
monthly = mkOption {
default = 12;
type = types.int;
description = ''
Number of monthly auto-snapshots that you wish to keep.
'';
};
};
};
2012-12-04 18:17:54 +00:00
###### implementation
config = mkMerge [
(mkIf enableZfs {
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
assertions = [
{
assertion = config.networking.hostId != null;
message = "ZFS requires config.networking.hostId to be set";
}
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
{
assertion = !cfgZfs.forceImportAll || cfgZfs.forceImportRoot;
message = "If you enable boot.zfs.forceImportAll, you must also enable boot.zfs.forceImportRoot";
}
];
boot = {
kernelModules = [ "spl" "zfs" ] ;
extraModulePackages = [ splKernelPkg zfsKernelPkg ];
};
boot.initrd = mkIf inInitrd {
kernelModules = [ "spl" "zfs" ];
extraUtilsCommands =
''
cp -v ${zfsUserPkg}/sbin/zfs $out/bin
cp -v ${zfsUserPkg}/sbin/zdb $out/bin
cp -v ${zfsUserPkg}/sbin/zpool $out/bin
cp -pdv ${zfsUserPkg}/lib/lib*.so* $out/lib
cp -pdv ${pkgs.zlib}/lib/lib*.so* $out/lib
'';
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
postDeviceCommands = concatStringsSep "\n" ([''
ZFS_FORCE="${optionalString cfgZfs.forceImportRoot "-f"}"
for o in $(cat /proc/cmdline); do
case $o in
zfs_force|zfs_force=1)
ZFS_FORCE="-f"
;;
esac
done
''] ++ (map (pool: ''
echo "importing root ZFS pool \"${pool}\"..."
zpool import -N $ZFS_FORCE "${pool}"
'') rootPools));
};
boot.loader.grub = mkIf inInitrd {
zfsSupport = true;
};
environment.etc."zfs/zed.d".source = "${zfsUserPkg}/etc/zfs/zed.d/*";
system.fsPackages = [ zfsUserPkg ]; # XXX: needed? zfs doesn't have (need) a fsck
environment.systemPackages = [ zfsUserPkg ];
services.udev.packages = [ zfsUserPkg ]; # to hook zvol naming, etc.
systemd.packages = [ zfsUserPkg ];
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
systemd.services = let
getPoolFilesystems = pool:
filter (x: x.fsType == "zfs" && (fsToPool x) == pool) (attrValues config.fileSystems);
getPoolMounts = pool:
let
mountPoint = fs: escapeSystemdPath fs.mountPoint;
in
map (x: "${mountPoint x}.mount") (getPoolFilesystems pool);
createImportService = pool:
nameValuePair "zfs-import-${pool}" {
description = "Import ZFS pool \"${pool}\"";
requires = [ "systemd-udev-settle.service" ];
after = [ "systemd-udev-settle.service" "systemd-modules-load.service" ];
wantedBy = (getPoolMounts pool) ++ [ "local-fs.target" ];
before = (getPoolMounts pool) ++ [ "local-fs.target" ];
unitConfig = {
DefaultDependencies = "no";
};
serviceConfig = {
Type = "oneshot";
RemainAfterExit = true;
};
script = ''
zpool_cmd="${zfsUserPkg}/sbin/zpool"
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
("$zpool_cmd" list "${pool}" >/dev/null) || "$zpool_cmd" import -N ${optionalString cfgZfs.forceImportAll "-f"} "${pool}"
'';
};
in listToAttrs (map createImportService dataPools) // {
"zfs-mount" = { after = [ "systemd-modules-load.service" ]; };
"zfs-share" = { after = [ "systemd-modules-load.service" ]; };
"zed" = { after = [ "systemd-modules-load.service" ]; };
};
systemd.targets."zfs-import" =
let
services = map (pool: "zfs-import-${pool}.service") dataPools;
in
{
requires = services;
after = services;
};
systemd.targets."zfs".wantedBy = [ "multi-user.target" ];
})
(mkIf enableAutoSnapshots {
systemd.services."zfs-snapshot-frequent" = {
description = "ZFS auto-snapshotting every 15 mins";
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
after = [ "zfs-import.target" ];
serviceConfig = {
Type = "oneshot";
ExecStart = "${zfsAutoSnap} frequent ${toString cfgSnapshots.frequent}";
};
restartIfChanged = false;
startAt = "*:15,30,45";
};
systemd.services."zfs-snapshot-hourly" = {
description = "ZFS auto-snapshotting every hour";
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
after = [ "zfs-import.target" ];
serviceConfig = {
Type = "oneshot";
ExecStart = "${zfsAutoSnap} hourly ${toString cfgSnapshots.hourly}";
};
restartIfChanged = false;
startAt = "hourly";
};
systemd.services."zfs-snapshot-daily" = {
description = "ZFS auto-snapshotting every day";
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
after = [ "zfs-import.target" ];
serviceConfig = {
Type = "oneshot";
ExecStart = "${zfsAutoSnap} daily ${toString cfgSnapshots.daily}";
};
restartIfChanged = false;
startAt = "daily";
};
systemd.services."zfs-snapshot-weekly" = {
description = "ZFS auto-snapshotting every week";
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
after = [ "zfs-import.target" ];
serviceConfig = {
Type = "oneshot";
ExecStart = "${zfsAutoSnap} weekly ${toString cfgSnapshots.weekly}";
};
restartIfChanged = false;
startAt = "weekly";
};
systemd.services."zfs-snapshot-monthly" = {
description = "ZFS auto-snapshotting every month";
nixos/zfs: Improve the ZFS boot process It turns out that the upstream systemd services that import ZFS pools contain serious bugs. The first major problem is that importing pools fails if there are no pools to import. The second major problem is that if a pool ends up in /etc/zfs/zpool.cache but it disappears from the system (e.g. if you reboot but during the reboot you unplug your ZFS-formatted USB pen drive), then the import service will always fail and it will be impossible to get rid of the pool from the cache (unless you manually delete the cache). Also, the upstream service would always import all available ZFS pools every boot, which may not be what is desired in some cases. This commit will solve these problems in the following ways: 1. Ignore /etc/zfs/zpool.cache. This seems to be a major source of issues, and also does not play well with NixOS's philosophy of reproducible configurations. Instead, on every boot NixOS will try to import the set of pools that are specified in its configuration. This is also the direction that upstream is moving towards. 2. Instead of trying to import all ZFS pools, only import those that are actually necessary. NixOS will automatically determine these from the config.fileSystems.* option. Also, the user can import any additional pools every boot by adding them to the config.boot.zfs.extraPools option, but this is only necessary if their filesystems are not specified in config.fileSystems.*. 3. Added options to configure if ZFS should force-import ZFS pools. This may currently be necessary, especially if your pools have not been correctly imported with a proper host id configuration (which is probably true for 99% of current NixOS ZFS users). Once host id configuration becomes mandatory when using ZFS in NixOS and we are sure that most users have updated their configurations and rebooted at least once, we should disable force-import by default. Probably, this shouldn't be done before the next stable release. WARNING: This commit may change the order in which your non-ZFS vs ZFS filesystems are mounted. To avoid this problem (now or in the future) it is recommended that you set the 'mountpoint' property of your ZFS filesystems to 'legacy', and that you manage them using config.fileSystems, just like any other non-ZFS filesystem is usually managed in NixOS.
2014-10-22 18:17:21 +01:00
after = [ "zfs-import.target" ];
serviceConfig = {
Type = "oneshot";
ExecStart = "${zfsAutoSnap} monthly ${toString cfgSnapshots.monthly}";
};
restartIfChanged = false;
startAt = "monthly";
};
})
];
2012-12-04 18:17:54 +00:00
}