Command Hooks
Run commands before, after success or after failure
resticprofile has 2 places where you can run commands around restic:
- commands that will run before and after every restic command (snapshots, backup, check, forget, prune, mount, etc.). These are placed at the root of each profile and are always considered.
- commands that will only run before and after specific restic commands. These are placed in supported sections of your profiles (currently supported are
backup
,copy
,dump
,find
,ls
,mount
,restore
,snapshots
,stats
andtag
).
Here’s an example of all the external commands that you can run during the execution of a profile:
version = "1"
[documents]
run-before = 'echo "== run-before profile $PROFILE_NAME and command $PROFILE_COMMAND"'
run-after = 'echo "== run-after profile $PROFILE_NAME and command $PROFILE_COMMAND"'
run-after-fail = 'echo "== ERROR in profile $PROFILE_NAME command ${PROFILE_COMMAND}: $ERROR_MESSAGE"'
run-finally = 'echo "== run-finally from profile $PROFILE_NAME after command $PROFILE_COMMAND"'
[documents.backup]
run-before = 'echo "=== run-before backup in profile $PROFILE_NAME"'
run-after = 'echo "=== run-after backup in profile $PROFILE_NAME"'
run-after-fail = 'echo "=== ERROR in backup for profile ${PROFILE_NAME}: $ERROR_MESSAGE"'
run-finally = 'echo "=== run-finally after backup in profile $PROFILE_NAME"'
source = "~/Documents"
version: "1"
documents:
run-before: 'echo "== run-before profile $PROFILE_NAME and command $PROFILE_COMMAND"'
run-after: 'echo "== run-after profile $PROFILE_NAME and command $PROFILE_COMMAND"'
run-after-fail: 'echo "== ERROR in profile $PROFILE_NAME command ${PROFILE_COMMAND}: $ERROR_MESSAGE"'
run-finally: 'echo "== run-finally from profile $PROFILE_NAME after command $PROFILE_COMMAND"'
backup:
run-before: 'echo "=== run-before backup in profile $PROFILE_NAME"'
run-after: 'echo "=== run-after backup in profile $PROFILE_NAME"'
run-after-fail: 'echo "=== ERROR in backup for profile ${PROFILE_NAME}: $ERROR_MESSAGE"'
run-finally: 'echo "=== run-finally after backup in profile $PROFILE_NAME"'
source: ~/Documents
documents {
run-before = "echo \"== run-before profile $PROFILE_NAME and command $PROFILE_COMMAND\""
run-after = "echo \"== run-after profile $PROFILE_NAME and command $PROFILE_COMMAND\""
run-after-fail = "echo \"== ERROR in profile $PROFILE_NAME command ${PROFILE_COMMAND}: $ERROR_MESSAGE\""
run-finally = "echo \"== run-finally from profile $PROFILE_NAME after command $PROFILE_COMMAND\""
backup = {
run-before = "echo \"=== run-before backup in profile $PROFILE_NAME\""
run-after = "echo \"=== run-after backup in profile $PROFILE_NAME\""
run-after-fail = "echo \"=== ERROR in backup for profile ${PROFILE_NAME}: $ERROR_MESSAGE\""
run-finally = "echo \"=== run-finally after backup in profile $PROFILE_NAME\""
source = "~/Documents"
}
}
{
"version": "1",
"documents": {
"run-before": "echo \"== run-before profile $PROFILE_NAME and command $PROFILE_COMMAND\"",
"run-after": "echo \"== run-after profile $PROFILE_NAME and command $PROFILE_COMMAND\"",
"run-after-fail": "echo \"== ERROR in profile $PROFILE_NAME command ${PROFILE_COMMAND}: $ERROR_MESSAGE\"",
"run-finally": "echo \"== run-finally from profile $PROFILE_NAME after command $PROFILE_COMMAND\"",
"backup": {
"run-before": "echo \"=== run-before backup in profile $PROFILE_NAME\"",
"run-after": "echo \"=== run-after backup in profile $PROFILE_NAME\"",
"run-after-fail": "echo \"=== ERROR in backup for profile ${PROFILE_NAME}: $ERROR_MESSAGE\"",
"run-finally": "echo \"=== run-finally after backup in profile $PROFILE_NAME\"",
"source": "~/Documents"
}
}
}
run-before
, run-after
, run-after-fail
and run-finally
can be a string, or an array of strings if you need to run more than one command
A few environment variables will be set before running these commands:
PROFILE_NAME
PROFILE_COMMAND
: backup, check, forget, etc.
Additionally, for the run-after-fail
commands, these environment variables will also be available:
ERROR_MESSAGE
(andERROR
) containing the latest error messageERROR_COMMANDLINE
containing the command line that failedERROR_EXIT_CODE
containing the exit code of the command line that failedERROR_STDERR
containing any message that the failed command sent to the standard error (stderr)
The commands of run-finally
get the environment of run-after-fail
when run-before
, run-after
or restic
failed.
Failures to run commands in run-finally
are logged but do not influence environment, return code nor running of further commands listed in run-finally
. This makes it the perfect choice for custom cleanup tasks that must always run.
All other command errors (= non-zero return code from a command) will skip running further commands in the same list and/or abort the flow.
Output when running the example from above
➜ resticprofile documents.backup
== run-before profile documents and command backup
=== run-before backup in profile documents
...
processed 355 files, 11.722 MiB in 0:00
snapshot 3949d2fb saved
...
=== run-after backup in profile documents
== run-after profile documents and command backup
=== run-finally after backup in profile documents
== run-finally from profile documents after command backup
Order of run-*
during a backup
The commands will be running in this order during a backup:
run-before
from the profile - if error, go torun-after-fail
run-before
from the backup section - if error, go torun-after-fail
from backup section- run the restic backup (with check and retention if configured) - if error, go to
run-after-fail
from backup section run-after
from the backup section - if error, go torun-after-fail
from backup sectionrun-after
from the profile - if error, go torun-after-fail
- If error:
run-after-fail
from the backup section - if error, go torun-finally
- If error:
run-after-fail
from the profile - if error, go torun-finally
run-finally
from the backup section - if error, log and continue with nextrun-finally
from the profile - if error, log and continue with next
Maybe it’s easier to understand with a flow diagram:
flowchart TB LOCK(set resticprofile lock) UNLOCK(delete resticprofile lock) PRB('run-before' from profile) PRA('run-after' from profile) subgraph Backup [ ] BRB('run-before' from backup section) BRA('run-after' from backup section) RUN(run restic backup with check and/or retention if configured) end subgraph Failure [ ] BFAIL('run-after-fail' from backup section) PFAIL('run-after-fail' from profile) end subgraph Finally [ ] direction TB BRF('run-finally' from backup section) PRF('run-finally' from profile) BRF --> PRF end LOCK --> PRB PRB -->|Error| PFAIL PRB -->|Success| BRB BRB -->|Error| BFAIL BRB -->|Success| RUN RUN -->|Error| BFAIL RUN -->|Success| BRA BRA -->|Error| BFAIL BRA -->|Success| PRA BFAIL -->|Error| Finally BFAIL --> PFAIL PRA -->|Error| PFAIL PRA -->|Success| Finally PFAIL --> Finally Finally --> UNLOCK style Backup fill:#9990,stroke:#9990 style Failure fill:#9990,stroke:#9990 style Finally fill:#9991,stroke:#9994,stroke-width:4px
The local resticprofile lock is surrounding the whole process. It means that the run-after-fail
target is not called if the lock cannot be obtained. This is a limitation of the current implementation.
Run commands on stream errors
In addition to hooks around profile and command execution, resticprofile allows to monitor the standard error stream of the current running command and trigger a custom hook when an output error line matches a regular expression pattern.
version = "1"
[default]
[[default.stream-error]]
pattern = ".+ERROR.+"
run = 'echo "Found ERROR at least 5 times in ${PROFILE_COMMAND}. Last matched line:"; cat - '
max-runs = 0
min-matches = 5
version: "1"
default:
stream-error:
- pattern: ".+ERROR.+"
run: 'echo "Found ERROR at least 5 times in ${PROFILE_COMMAND}. Last matched line:"; cat - '
max-runs: 0
min-matches: 5
default {
stream-error = {
pattern = ".+ERROR.+"
run = "echo \"Found ERROR at least 5 times in ${PROFILE_COMMAND}. Last matched line:\"; cat - "
max-runs = 0
min-matches = 5
}
}
{
"version": "1",
"default": {
"stream-error": [
{
"pattern": ".+ERROR.+",
"run": "echo \"Found ERROR at least 5 times in ${PROFILE_COMMAND}. Last matched line:\"; cat - ",
"max-runs": 0,
"min-matches": 5
}
]
}
}
The config structure stream-error
can be defined multiple times within a profile and consists of the following properties:
pattern
: a regular expression pattern matched against every line of stderr.run
: the command to run whenpattern
is found in stderr.max-runs
: maximum number of times thatrun
is started for a single monitored command (or 0 for no limit).min-matches
: minimum number of times thatpattern
must match before startingrun
.
The run
command inherits the environment of the monitored command on a pattern match, which always includes:
PROFILE_NAME
PROFILE_COMMAND
: backup, check, forget, etc.
Stdin of the run
command is populated with the error output line that matched the pattern
. Using "cat -"
in run
effectively prints the line that triggered the action.
The following example shows how this could have been used with restic to address check
failures caused by over usage of /tmp/
(restic fixed this problem in 0.14.0):
version = "1"
[default]
[[default.stream-error]]
pattern = "/tmp/restic-check-cache.+no space left on device"
run = "rm -Rf /tmp/restic-check-cache*"
[[default.stream-error]]
pattern = "mkdir /tmp/restic-check-cache.+no such file or directory"
run = "cat - | cut -d : -f 2 - | grep -E 'mkdir /tmp[^ \\.]+$' | sed 's/mkdir/mkdir -p/' - | sh"
version: "1"
default:
stream-error:
- pattern: "/tmp/restic-check-cache.+no space left on device"
run: "rm -Rf /tmp/restic-check-cache*"
- pattern: "mkdir /tmp/restic-check-cache.+no such file or directory"
run: "cat - | cut -d : -f 2 - | grep -E 'mkdir /tmp[^ \\.]+$' | sed 's/mkdir/mkdir -p/' - | sh"