Mini Shell
"""
States for managing zfs datasets
:maintainer: Jorge Schrauwen <sjorge@blackdot.be>
:maturity: new
:depends: salt.utils.zfs, salt.modules.zfs
:platform: smartos, illumos, solaris, freebsd, linux
.. versionadded:: 2016.3.0
.. versionchanged:: 2018.3.1
Big refactor to remove duplicate code, better type conversions and improved
consistency in output.
.. code-block:: yaml
test/shares/yuki:
zfs.filesystem_present:
- create_parent: true
- properties:
quota: 16G
test/iscsi/haruhi:
zfs.volume_present:
- create_parent: true
- volume_size: 16M
- sparse: true
- properties:
readonly: on
test/shares/yuki@frozen:
zfs.snapshot_present
moka_origin:
zfs.hold_present:
- snapshot: test/shares/yuki@frozen
test/shares/moka:
zfs.filesystem_present:
- cloned_from: test/shares/yuki@frozen
test/shares/moka@tsukune:
zfs.snapshot_absent
"""
import logging
from datetime import datetime
from salt.utils.odict import OrderedDict
log = logging.getLogger(__name__)
# Define the state's virtual name
__virtualname__ = "zfs"
# Compare modifiers for zfs.schedule_snapshot
comp_hour = {"minute": 0}
comp_day = {"minute": 0, "hour": 0}
comp_month = {"minute": 0, "hour": 0, "day": 1}
comp_year = {"minute": 0, "hour": 0, "day": 1, "month": 1}
def __virtual__():
"""
Provides zfs state
"""
if not __grains__.get("zfs_support"):
return False, "The zfs state cannot be loaded: zfs not supported"
return __virtualname__
def _absent(name, dataset_type, force=False, recursive=False):
"""
internal shared function for *_absent
name : string
name of dataset
dataset_type : string [filesystem, volume, snapshot, or bookmark]
type of dataset to remove
force : boolean
try harder to destroy the dataset
recursive : boolean
also destroy all the child datasets
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## log configuration
dataset_type = dataset_type.lower()
log.debug("zfs.%s_absent::%s::config::force = %s", dataset_type, name, force)
log.debug(
"zfs.%s_absent::%s::config::recursive = %s", dataset_type, name, recursive
)
## destroy dataset if needed
if __salt__["zfs.exists"](name, **{"type": dataset_type}):
## NOTE: dataset found with the name and dataset_type
if not __opts__["test"]:
mod_res = __salt__["zfs.destroy"](
name, **{"force": force, "recursive": recursive}
)
else:
mod_res = OrderedDict([("destroyed", True)])
ret["result"] = mod_res["destroyed"]
if ret["result"]:
ret["changes"][name] = "destroyed"
ret["comment"] = "{} {} was destroyed".format(
dataset_type,
name,
)
else:
ret["comment"] = "failed to destroy {} {}".format(
dataset_type,
name,
)
if "error" in mod_res:
ret["comment"] = mod_res["error"]
else:
## NOTE: no dataset found with name of the dataset_type
ret["comment"] = f"{dataset_type} {name} is absent"
return ret
def filesystem_absent(name, force=False, recursive=False):
"""
ensure filesystem is absent on the system
name : string
name of filesystem
force : boolean
try harder to destroy the dataset (zfs destroy -f)
recursive : boolean
also destroy all the child datasets (zfs destroy -r)
.. warning::
If a volume with ``name`` exists, this state will succeed without
destroying the volume specified by ``name``. This module is dataset type sensitive.
"""
if not __utils__["zfs.is_dataset"](name):
ret = {
"name": name,
"changes": {},
"result": False,
"comment": f"invalid dataset name: {name}",
}
else:
ret = _absent(name, "filesystem", force, recursive)
return ret
def volume_absent(name, force=False, recursive=False):
"""
ensure volume is absent on the system
name : string
name of volume
force : boolean
try harder to destroy the dataset (zfs destroy -f)
recursive : boolean
also destroy all the child datasets (zfs destroy -r)
.. warning::
If a filesystem with ``name`` exists, this state will succeed without
destroying the filesystem specified by ``name``. This module is dataset type sensitive.
"""
if not __utils__["zfs.is_dataset"](name):
ret = {
"name": name,
"changes": {},
"result": False,
"comment": f"invalid dataset name: {name}",
}
else:
ret = _absent(name, "volume", force, recursive)
return ret
def snapshot_absent(name, force=False, recursive=False):
"""
ensure snapshot is absent on the system
name : string
name of snapshot
force : boolean
try harder to destroy the dataset (zfs destroy -f)
recursive : boolean
also destroy all the child datasets (zfs destroy -r)
"""
if not __utils__["zfs.is_snapshot"](name):
ret = {
"name": name,
"changes": {},
"result": False,
"comment": f"invalid snapshot name: {name}",
}
else:
ret = _absent(name, "snapshot", force, recursive)
return ret
def bookmark_absent(name, force=False, recursive=False):
"""
ensure bookmark is absent on the system
name : string
name of snapshot
force : boolean
try harder to destroy the dataset (zfs destroy -f)
recursive : boolean
also destroy all the child datasets (zfs destroy -r)
"""
if not __utils__["zfs.is_bookmark"](name):
ret = {
"name": name,
"changes": {},
"result": False,
"comment": f"invalid bookmark name: {name}",
}
else:
ret = _absent(name, "bookmark", force, recursive)
return ret
def hold_absent(name, snapshot, recursive=False):
"""
ensure hold is absent on the system
name : string
name of hold
snapshot : string
name of snapshot
recursive : boolean
recursively releases a hold with the given tag on the snapshots of all descendent file systems.
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## log configuration
log.debug("zfs.hold_absent::%s::config::snapshot = %s", name, snapshot)
log.debug("zfs.hold_absent::%s::config::recursive = %s", name, recursive)
## check we have a snapshot/tag name
if not __utils__["zfs.is_snapshot"](snapshot):
ret["result"] = False
ret["comment"] = f"invalid snapshot name: {snapshot}"
return ret
if (
__utils__["zfs.is_snapshot"](name)
or __utils__["zfs.is_bookmark"](name)
or name == "error"
):
ret["result"] = False
ret["comment"] = f"invalid tag name: {name}"
return ret
## release hold if required
holds = __salt__["zfs.holds"](snapshot)
if name in holds:
## NOTE: hold found for snapshot, release it
if not __opts__["test"]:
mod_res = __salt__["zfs.release"](
name, snapshot, **{"recursive": recursive}
)
else:
mod_res = OrderedDict([("released", True)])
ret["result"] = mod_res["released"]
if ret["result"]:
ret["changes"] = {snapshot: {name: "released"}}
ret["comment"] = "hold {} released".format(
name,
)
else:
ret["comment"] = "failed to release hold {}".format(
name,
)
if "error" in mod_res:
ret["comment"] = mod_res["error"]
elif "error" in holds:
## NOTE: we have an error
ret["result"] = False
ret["comment"] = holds["error"]
else:
## NOTE: no hold found with name for snapshot
ret["comment"] = "hold {} is absent".format(
name,
)
return ret
def hold_present(name, snapshot, recursive=False):
"""
ensure hold is present on the system
name : string
name of holdt
snapshot : string
name of snapshot
recursive : boolean
recursively add hold with the given tag on the snapshots of all descendent file systems.
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## log configuration
log.debug("zfs.hold_present::%s::config::snapshot = %s", name, snapshot)
log.debug("zfs.hold_present::%s::config::recursive = %s", name, recursive)
## check we have a snapshot/tag name
if not __utils__["zfs.is_snapshot"](snapshot):
ret["result"] = False
ret["comment"] = f"invalid snapshot name: {snapshot}"
return ret
if (
__utils__["zfs.is_snapshot"](name)
or __utils__["zfs.is_bookmark"](name)
or name == "error"
):
ret["result"] = False
ret["comment"] = f"invalid tag name: {name}"
return ret
## place hold if required
holds = __salt__["zfs.holds"](snapshot)
if name in holds:
## NOTE: hold with name already exists for snapshot
ret["comment"] = "hold {} is present for {}".format(
name,
snapshot,
)
else:
## NOTE: no hold found with name for snapshot
if not __opts__["test"]:
mod_res = __salt__["zfs.hold"](name, snapshot, **{"recursive": recursive})
else:
mod_res = OrderedDict([("held", True)])
ret["result"] = mod_res["held"]
if ret["result"]:
ret["changes"] = OrderedDict([(snapshot, OrderedDict([(name, "held")]))])
ret["comment"] = f"hold {name} added to {snapshot}"
else:
ret["comment"] = f"failed to add hold {name} to {snapshot}"
if "error" in mod_res:
ret["comment"] = mod_res["error"]
return ret
def _dataset_present(
dataset_type,
name,
properties,
volume_size=None,
sparse=False,
create_parent=False,
cloned_from=None,
):
"""
internal handler for filesystem_present/volume_present
dataset_type : string
volume or filesystem
name : string
name of volume
volume_size : string
size of volume
sparse : boolean
create sparse volume
create_parent : boolean
creates all the non-existing parent datasets.
any property specified on the command line using the -o option is ignored.
cloned_from : string
name of snapshot to clone
properties : dict
additional zfs properties (-o)
.. note::
``cloned_from`` is only use if the volume does not exist yet,
when ``cloned_from`` is set after the volume exists it will be ignored.
.. note::
Properties do not get cloned, if you specify the properties in the state file
they will be applied on a subsequent run.
``volume_size`` is considered a property, so the volume's size will be
corrected when the properties get updated if it differs from the
original volume.
The sparse parameter is ignored when using ``cloned_from``.
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## fallback dataset_type to filesystem if out of range
if dataset_type not in ["filesystem", "volume"]:
dataset_type = "filesystem"
## ensure properties are zfs values
if properties is None:
properties = {}
properties = __utils__["zfs.from_auto_dict"](properties)
if volume_size:
## NOTE: add volsize to properties
volume_size = __utils__["zfs.from_size"](volume_size)
properties.update({"volsize": volume_size})
# The sorted isn't necessary for proper behavior, but it helps for the unit
# tests.
propnames = ",".join(sorted(properties.keys()))
## log configuration
log.debug(
"zfs.%s_present::%s::config::volume_size = %s", dataset_type, name, volume_size
)
log.debug("zfs.%s_present::%s::config::sparse = %s", dataset_type, name, sparse)
log.debug(
"zfs.%s_present::%s::config::create_parent = %s",
dataset_type,
name,
create_parent,
)
log.debug(
"zfs.%s_present::%s::config::cloned_from = %s", dataset_type, name, cloned_from
)
log.debug(
"zfs.%s_present::%s::config::properties = %s", dataset_type, name, properties
)
## check we have valid filesystem name/volume name/clone snapshot
if not __utils__["zfs.is_dataset"](name):
ret["result"] = False
ret["comment"] = f"invalid dataset name: {name}"
return ret
if cloned_from and not __utils__["zfs.is_snapshot"](cloned_from):
ret["result"] = False
ret["comment"] = f"{cloned_from} is not a snapshot"
return ret
## ensure dataset is in correct state
## NOTE: update the dataset
exists = __salt__["zfs.exists"](name, **{"type": dataset_type})
if exists and len(properties) == 0:
ret["comment"] = f"{dataset_type} {name} is uptodate"
elif exists and len(properties) > 0:
## NOTE: fetch current volume properties
properties_current = __salt__["zfs.get"](
name,
properties=propnames,
type=dataset_type,
fields="value",
depth=0,
parsable=True,
).get(name, OrderedDict())
## NOTE: build list of properties to update
properties_update = []
for prop in properties:
## NOTE: skip unexisting properties
if prop not in properties_current:
log.warning(
"zfs.%s_present::%s::update - unknown property: %s",
dataset_type,
name,
prop,
)
continue
## NOTE: compare current and wanted value
if properties_current[prop]["value"] != properties[prop]:
properties_update.append(prop)
## NOTE: update pool properties
for prop in properties_update:
if not __opts__["test"]:
mod_res = __salt__["zfs.set"](name, **{prop: properties[prop]})
else:
mod_res = OrderedDict([("set", True)])
if mod_res["set"]:
if name not in ret["changes"]:
ret["changes"][name] = {}
ret["changes"][name][prop] = properties[prop]
else:
ret["result"] = False
if ret["comment"] == "":
ret["comment"] = "The following properties were not updated:"
ret["comment"] = "{} {}".format(ret["comment"], prop)
## NOTE: update comment
if ret["result"] and name in ret["changes"]:
ret["comment"] = f"{dataset_type} {name} was updated"
elif ret["result"]:
ret["comment"] = f"{dataset_type} {name} is uptodate"
else:
ret["comment"] = f"{dataset_type} {name} failed to be updated"
## NOTE: create or clone the dataset
elif not exists:
mod_res_action = "cloned" if cloned_from else "created"
if __opts__["test"]:
## NOTE: pretend to create/clone
mod_res = OrderedDict([(mod_res_action, True)])
elif cloned_from:
## NOTE: add volsize to properties
if volume_size:
properties["volsize"] = volume_size
## NOTE: clone the dataset
mod_res = __salt__["zfs.clone"](
cloned_from,
name,
**{"create_parent": create_parent, "properties": properties},
)
else:
## NOTE: create the dataset
mod_res = __salt__["zfs.create"](
name,
**{
"create_parent": create_parent,
"properties": properties,
"volume_size": volume_size,
"sparse": sparse,
},
)
ret["result"] = mod_res[mod_res_action]
if ret["result"]:
ret["changes"][name] = mod_res_action
if properties:
ret["changes"][name] = properties
ret["comment"] = "{} {} was {}".format(
dataset_type,
name,
mod_res_action,
)
else:
ret["comment"] = "failed to {} {} {}".format(
mod_res_action[:-1],
dataset_type,
name,
)
if "error" in mod_res:
ret["comment"] = mod_res["error"]
return ret
def filesystem_present(name, create_parent=False, properties=None, cloned_from=None):
"""
ensure filesystem exists and has properties set
name : string
name of filesystem
create_parent : boolean
creates all the non-existing parent datasets.
any property specified on the command line using the -o option is ignored.
cloned_from : string
name of snapshot to clone
properties : dict
additional zfs properties (-o)
.. note::
``cloned_from`` is only use if the filesystem does not exist yet,
when ``cloned_from`` is set after the filesystem exists it will be ignored.
.. note::
Properties do not get cloned, if you specify the properties in the
state file they will be applied on a subsequent run.
"""
return _dataset_present(
"filesystem",
name,
properties,
create_parent=create_parent,
cloned_from=cloned_from,
)
def volume_present(
name,
volume_size,
sparse=False,
create_parent=False,
properties=None,
cloned_from=None,
):
"""
ensure volume exists and has properties set
name : string
name of volume
volume_size : string
size of volume
sparse : boolean
create sparse volume
create_parent : boolean
creates all the non-existing parent datasets.
any property specified on the command line using the -o option is ignored.
cloned_from : string
name of snapshot to clone
properties : dict
additional zfs properties (-o)
.. note::
``cloned_from`` is only use if the volume does not exist yet,
when ``cloned_from`` is set after the volume exists it will be ignored.
.. note::
Properties do not get cloned, if you specify the properties in the state file
they will be applied on a subsequent run.
``volume_size`` is considered a property, so the volume's size will be
corrected when the properties get updated if it differs from the
original volume.
The sparse parameter is ignored when using ``cloned_from``.
"""
return _dataset_present(
"volume",
name,
properties,
volume_size,
sparse=sparse,
create_parent=create_parent,
cloned_from=cloned_from,
)
def bookmark_present(name, snapshot):
"""
ensure bookmark exists
name : string
name of bookmark
snapshot : string
name of snapshot
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## log configuration
log.debug("zfs.bookmark_present::%s::config::snapshot = %s", name, snapshot)
## check we have valid snapshot/bookmark name
if not __utils__["zfs.is_snapshot"](snapshot):
ret["result"] = False
ret["comment"] = f"invalid snapshot name: {name}"
return ret
if "#" not in name and "/" not in name:
## NOTE: simple snapshot name
# take the snapshot name and replace the snapshot but with the simple name
# e.g. pool/fs@snap + bm --> pool/fs#bm
name = "{}#{}".format(snapshot[: snapshot.index("@")], name)
ret["name"] = name
if not __utils__["zfs.is_bookmark"](name):
ret["result"] = False
ret["comment"] = f"invalid bookmark name: {name}"
return ret
## ensure bookmark exists
if not __salt__["zfs.exists"](name, **{"type": "bookmark"}):
## NOTE: bookmark the snapshot
if not __opts__["test"]:
mod_res = __salt__["zfs.bookmark"](snapshot, name)
else:
mod_res = OrderedDict([("bookmarked", True)])
ret["result"] = mod_res["bookmarked"]
if ret["result"]:
ret["changes"][name] = snapshot
ret["comment"] = f"{snapshot} bookmarked as {name}"
else:
ret["comment"] = f"failed to bookmark {snapshot}"
if "error" in mod_res:
ret["comment"] = mod_res["error"]
else:
## NOTE: bookmark already exists
ret["comment"] = "bookmark is present"
return ret
def snapshot_present(name, recursive=False, properties=None):
"""
ensure snapshot exists and has properties set
name : string
name of snapshot
recursive : boolean
recursively create snapshots of all descendent datasets
properties : dict
additional zfs properties (-o)
.. note:
Properties are only set at creation time
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## log configuration
log.debug("zfs.snapshot_present::%s::config::recursive = %s", name, recursive)
log.debug("zfs.snapshot_present::%s::config::properties = %s", name, properties)
## ensure properties are zfs values
if properties:
properties = __utils__["zfs.from_auto_dict"](properties)
## check we have valid snapshot name
if not __utils__["zfs.is_snapshot"](name):
ret["result"] = False
ret["comment"] = f"invalid snapshot name: {name}"
return ret
## ensure snapshot exits
if not __salt__["zfs.exists"](name, **{"type": "snapshot"}):
## NOTE: create the snapshot
if not __opts__["test"]:
mod_res = __salt__["zfs.snapshot"](
name, **{"recursive": recursive, "properties": properties}
)
else:
mod_res = OrderedDict([("snapshotted", True)])
ret["result"] = mod_res["snapshotted"]
if ret["result"]:
ret["changes"][name] = "snapshotted"
if properties:
ret["changes"][name] = properties
ret["comment"] = f"snapshot {name} was created"
else:
ret["comment"] = f"failed to create snapshot {name}"
if "error" in mod_res:
ret["comment"] = mod_res["error"]
else:
## NOTE: snapshot already exists
ret["comment"] = "snapshot is present"
return ret
def promoted(name):
"""
ensure a dataset is not a clone
name : string
name of fileset or volume
.. warning::
only one dataset can be the origin,
if you promote a clone the original will now point to the promoted dataset
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## check we if we have a valid dataset name
if not __utils__["zfs.is_dataset"](name):
ret["result"] = False
ret["comment"] = f"invalid dataset name: {name}"
return ret
## ensure dataset is the primary instance
if not __salt__["zfs.exists"](name, **{"type": "filesystem,volume"}):
## NOTE: we don't have a dataset
ret["result"] = False
ret["comment"] = f"dataset {name} does not exist"
else:
## NOTE: check if we have a blank origin (-)
if (
__salt__["zfs.get"](
name, **{"properties": "origin", "fields": "value", "parsable": True}
)[name]["origin"]["value"]
== "-"
):
## NOTE: we're already promoted
ret["comment"] = f"{name} already promoted"
else:
## NOTE: promote dataset
if not __opts__["test"]:
mod_res = __salt__["zfs.promote"](name)
else:
mod_res = OrderedDict([("promoted", True)])
ret["result"] = mod_res["promoted"]
if ret["result"]:
ret["changes"][name] = "promoted"
ret["comment"] = f"{name} promoted"
else:
ret["comment"] = f"failed to promote {name}"
if "error" in mod_res:
ret["comment"] = mod_res["error"]
return ret
def _schedule_snapshot_retrieve(dataset, prefix, snapshots):
"""
Update snapshots dict with current snapshots
dataset: string
name of filesystem or volume
prefix : string
prefix for the snapshots
e.g. 'test' will result in snapshots being named 'test-yyyymmdd_hhmm'
snapshots : OrderedDict
preseeded OrderedDict with configuration
"""
## NOTE: retrieve all snapshots for the dataset
for snap in sorted(
__salt__["zfs.list"](
dataset, **{"recursive": True, "depth": 1, "type": "snapshot"}
).keys()
):
## NOTE: we only want the actualy name
## myzpool/data@zbck-20171201_000248 -> zbck-20171201_000248
snap_name = snap[snap.index("@") + 1 :]
## NOTE: we only want snapshots matching our prefix
if not snap_name.startswith(f"{prefix}-"):
continue
## NOTE: retrieve the holds for this snapshot
snap_holds = __salt__["zfs.holds"](snap)
## NOTE: this snapshot has no holds, eligable for pruning
if not snap_holds:
snapshots["_prunable"].append(snap)
## NOTE: update snapshots based on holds (if any)
## we are only interested in the ones from our schedule
## if we find any others we skip them
for hold in snap_holds:
if hold in snapshots["_schedule"].keys():
snapshots[hold].append(snap)
return snapshots
def _schedule_snapshot_prepare(dataset, prefix, snapshots):
"""
Update snapshots dict with info for a new snapshot
dataset: string
name of filesystem or volume
prefix : string
prefix for the snapshots
e.g. 'test' will result in snapshots being named 'test-yyyymmdd_hhmm'
snapshots : OrderedDict
preseeded OrderedDict with configuration
"""
## NOTE: generate new snapshot name
snapshot_create_name = "{dataset}@{prefix}-{timestamp}".format(
dataset=dataset,
prefix=prefix,
timestamp=datetime.now().strftime("%Y%m%d_%H%M%S"),
)
## NOTE: figure out if we need to create the snapshot
timestamp_now = datetime.now().replace(second=0, microsecond=0)
snapshots["_create"][snapshot_create_name] = []
for hold, hold_count in snapshots["_schedule"].items():
## NOTE: skip hold if we don't keep snapshots for it
if hold_count == 0:
continue
## NOTE: figure out if we need the current hold on the new snapshot
if snapshots[hold]:
## NOTE: extract datetime from snapshot name
timestamp = datetime.strptime(
snapshots[hold][-1],
f"{dataset}@{prefix}-%Y%m%d_%H%M%S",
).replace(second=0, microsecond=0)
## NOTE: compare current timestamp to timestamp from snapshot
if hold == "minute" and timestamp_now <= timestamp:
continue
elif hold == "hour" and timestamp_now.replace(
**comp_hour
) <= timestamp.replace(**comp_hour):
continue
elif hold == "day" and timestamp_now.replace(
**comp_day
) <= timestamp.replace(**comp_day):
continue
elif hold == "month" and timestamp_now.replace(
**comp_month
) <= timestamp.replace(**comp_month):
continue
elif hold == "year" and timestamp_now.replace(
**comp_year
) <= timestamp.replace(**comp_year):
continue
## NOTE: add hold entry for snapshot
snapshots["_create"][snapshot_create_name].append(hold)
return snapshots
def scheduled_snapshot(name, prefix, recursive=True, schedule=None):
"""
maintain a set of snapshots based on a schedule
name : string
name of filesystem or volume
prefix : string
prefix for the snapshots
e.g. 'test' will result in snapshots being named 'test-yyyymmdd_hhmm'
recursive : boolean
create snapshots for all children also
schedule : dict
dict holding the schedule, the following keys are available (minute, hour,
day, month, and year) by default all are set to 0 the value indicated the
number of snapshots of that type to keep around.
.. warning::
snapshots will only be created and pruned every time the state runs.
a schedule must be setup to automatically run the state. this means that if
you run the state daily the hourly snapshot will only be made once per day!
.. versionchanged:: 2018.3.0
switched to localtime from gmtime so times now take into account timezones.
"""
ret = {"name": name, "changes": {}, "result": True, "comment": ""}
## initialize defaults
schedule_holds = ["minute", "hour", "day", "month", "year"]
snapshots = OrderedDict(
[("_create", OrderedDict()), ("_prunable", []), ("_schedule", OrderedDict())]
)
## strict configuration validation
## NOTE: we need a valid dataset
if not __utils__["zfs.is_dataset"](name):
ret["result"] = False
ret["comment"] = f"invalid dataset name: {name}"
if not __salt__["zfs.exists"](name, **{"type": "filesystem,volume"}):
ret["comment"] = f"dataset {name} does not exist"
ret["result"] = False
## NOTE: prefix must be 4 or longer
if not prefix or len(prefix) < 4:
ret["comment"] = f"prefix ({prefix}) must be at least 4 long"
ret["result"] = False
## NOTE: validate schedule
total_count = 0
for hold in schedule_holds:
snapshots[hold] = []
if hold not in schedule:
snapshots["_schedule"][hold] = 0
elif isinstance(schedule[hold], int):
snapshots["_schedule"][hold] = schedule[hold]
else:
ret["result"] = False
ret["comment"] = "schedule value for {} is not an integer".format(
hold,
)
break
total_count += snapshots["_schedule"][hold]
if ret["result"] and total_count == 0:
ret["result"] = False
ret["comment"] = "schedule is not valid, you need to keep atleast 1 snapshot"
## NOTE: return if configuration is not valid
if not ret["result"]:
return ret
## retrieve existing snapshots
snapshots = _schedule_snapshot_retrieve(name, prefix, snapshots)
## prepare snapshot
snapshots = _schedule_snapshot_prepare(name, prefix, snapshots)
## log configuration
log.debug("zfs.scheduled_snapshot::%s::config::recursive = %s", name, recursive)
log.debug("zfs.scheduled_snapshot::%s::config::prefix = %s", name, prefix)
log.debug("zfs.scheduled_snapshot::%s::snapshots = %s", name, snapshots)
## create snapshot(s)
for snapshot_name, snapshot_holds in snapshots["_create"].items():
## NOTE: skip if new snapshot has no holds
if not snapshot_holds:
continue
## NOTE: create snapshot
if not __opts__["test"]:
mod_res = __salt__["zfs.snapshot"](
snapshot_name, **{"recursive": recursive}
)
else:
mod_res = OrderedDict([("snapshotted", True)])
if not mod_res["snapshotted"]:
ret["result"] = False
ret["comment"] = f"error creating snapshot ({snapshot_name})"
else:
## NOTE: create holds (if we have a snapshot)
for hold in snapshot_holds:
if not __opts__["test"]:
mod_res = __salt__["zfs.hold"](
hold, snapshot_name, **{"recursive": recursive}
)
else:
mod_res = OrderedDict([("held", True)])
if not mod_res["held"]:
ret["result"] = False
ret["comment"] = "error adding hold ({}) to snapshot ({})".format(
hold,
snapshot_name,
)
break
snapshots[hold].append(snapshot_name)
if ret["result"]:
ret["comment"] = "scheduled snapshots updated"
if "created" not in ret["changes"]:
ret["changes"]["created"] = []
ret["changes"]["created"].append(snapshot_name)
## prune hold(s)
for hold, hold_count in snapshots["_schedule"].items():
while ret["result"] and len(snapshots[hold]) > hold_count:
## NOTE: pop oldest snapshot
snapshot_name = snapshots[hold].pop(0)
## NOTE: release hold for snapshot
if not __opts__["test"]:
mod_res = __salt__["zfs.release"](
hold, snapshot_name, **{"recursive": recursive}
)
else:
mod_res = OrderedDict([("released", True)])
if not mod_res["released"]:
ret["result"] = False
ret["comment"] = "error adding hold ({}) to snapshot ({})".format(
hold,
snapshot_name,
)
## NOTE: mark as prunable
if not __salt__["zfs.holds"](snapshot_name):
snapshots["_prunable"].append(snapshot_name)
## prune snapshot(s)
for snapshot_name in snapshots["_prunable"]:
## NOTE: destroy snapshot
if not __opts__["test"]:
mod_res = __salt__["zfs.destroy"](snapshot_name, **{"recursive": recursive})
else:
mod_res = OrderedDict([("destroyed", True)])
if not mod_res["destroyed"]:
ret["result"] = False
ret["comment"] = "error prunding snapshot ({1})".format(
snapshot_name,
)
break
if ret["result"] and snapshots["_prunable"]:
ret["comment"] = "scheduled snapshots updated"
ret["changes"]["pruned"] = snapshots["_prunable"]
if ret["result"] and not ret["changes"]:
ret["comment"] = "scheduled snapshots are up to date"
return ret
Zerion Mini Shell 1.0