Merge ~dmzoneill/charm-hw-health:bug/1901735 into charm-hw-health:master

[David O Neill]

[James Troup (elmo)]

Please use ISO 8601 format for the data, i.e. YYYY-MM-DD as that is universal and unambiguous

review: Needs Fixing

[David O Neill (dmzoneill)]

root@JFA1-0502-Compute-4:/home/ubuntu# ipmi-sel --help | grep date-range
      --date-range=MM/DD/YYYY-MM/DD/YYYY
      --exclude-date-range=MM/DD/YYYY-MM/DD/YYYY

[Linda Guo (lihuiguo)]

see the inline comments, needs to check if '--seloptions' already exists in cron job command line options.

[Xav Paice (xavpaice)]

this was reworked and merged in https://code.launchpad.net/~ziyiwang/charm-hw-health/+git/charm-hw-health/+merge/400599

Unmerged commits

3b2dfc2... by David O Neill on 2021-02-01

LP1901735 - data filtering

The ipmi-sel log can be filtered by date.
You must first acknowledge the previous ipmi sel entries by either specifying a date
or no date.
When specifying no date, that will default to the current date.

juju run-action hw-health/8 ack-sel --wait
juju run-action hw-health/8 ack-sel date=02/23/2019 --wait
juju run-action hw-health/8 show-sel --wait

diff --git a/src/README.md b/src/README.md
index e367400..fdba28c 100644
--- a/src/README.md
+++ b/src/README.md
@@ -54,6 +54,20 @@ zip /tmp/zipfile.zip megacli
 etc.
 ```
 
+# Usage continued
+
+The ipmi-sel log can be filtered by date.
+You must first acknowledge the previous ipmi sel entries by either specifying a date
+or no date.
+When specifying no date, that will default to the current date.
+```
+juju run-action hw-health/8 ack-sel --wait
+# or
+juju run-action hw-health/8 ack-sel date=02/23/2019 --wait
+# view
+juju run-action hw-health/8 show-sel --wait
+```
+
 ## Known Limitations and Issues
 
 Charm only install method is via Juju resources. There are plans to support
diff --git a/src/actions.yaml b/src/actions.yaml
index 2e691b8..dbc4553 100644
--- a/src/actions.yaml
+++ b/src/actions.yaml
@@ -10,3 +10,10 @@ show-sel:
     show-all:
       type: boolean
       description: Show all events, not just the non-nominal ones.
+ack-sel:
+  description: |
+    Set date filter for ipmi-sel to filter out old logs from ongoing reports.
+  params:
+    date:
+      type: string
+      description: From what date to filter (default now if unspecified)
diff --git a/src/actions/ack-sel b/src/actions/ack-sel
 new file mode 120000
index 0000000..405a394
--- /dev/null
+++ b/src/actions/ack-sel
@@ -0,0 +1 @@
+actions.py
 \ No newline at end of file
diff --git a/src/actions/actions.py b/src/actions/actions.py
index 9a84c5f..e4b2b65 100755
--- a/src/actions/actions.py
+++ b/src/actions/actions.py
@@ -14,13 +14,24 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Define hw-health charm actions."""
-
 import os
 import subprocess
 import sys
+from datetime import datetime, timedelta
+from traceback import format_exc
+
+# Load modules from $CHARM_DIR/lib
+sys.path.append("lib")
+
+import charmhelpers.core.hookenv as hookenv  # noqa: E402
+import charmhelpers.core.unitdata as unitdata  # noqa: E402
 
-from charmhelpers.core.hookenv import action_fail, action_get, action_set, log
+import charms.reactive  # noqa: E402
+from charms.layer import basic  # noqa: E402
+from charms.reactive.flags import clear_flag  # noqa: E402
 
+basic.bootstrap_charm_deps()
+basic.init_config_states()
 
 IPMI_SEL = "/usr/sbin/ipmi-sel"
 
@@ -34,10 +45,10 @@ def clear_sel():
     command = [IPMI_SEL, "--post-clear"]
     try:
         output = subprocess.check_output(command)
-        log("Action clear-sel completed, sel log cleared: {}".format(output))
-        action_set({"message": output.decode("UTF-8")})
+        hookenv.log("Action clear-sel completed, sel log cleared: {}".format(output))
+        hookenv.action_set({"message": output.decode("UTF-8")})
     except subprocess.CalledProcessError as e:
-        action_fail("Action failed with {}".format(e))
+        hookenv.action_fail("Action failed with {}".format(e))
 
 
 def show_sel():
@@ -46,8 +57,20 @@ def show_sel():
     By default, this will show all non-nominal events.
     If you specify show-all, it will show all events.
     """
-    show_all = action_get("show-all")
-    command = [IPMI_SEL, "--output-event-state"]
+    command = None
+    show_all = hookenv.action_get("show-all")
+    date_f = unitdata.kv().get("date_filter")
+    if not date_f:
+        command = [IPMI_SEL, "--output-event-state"]
+    else:
+        # if you ack-sel without specifying a date
+        # the ack-sel default to now()
+        # current date -> now is invalid date range
+        # now -> tomorrow
+        tomorrow = datetime.now() + timedelta(days=1)
+        drange = str(date_f) + "-" + tomorrow.strftime("%m/%d/%Y")
+        command = [IPMI_SEL, "--output-event-state", "--date-range", drange]
+
     try:
         header, body = None, None
         output = subprocess.check_output(command).decode("UTF-8")
@@ -59,12 +82,36 @@ def show_sel():
                 body = [line for line in body if "Nominal" not in line]
         if body:
             final_output = "\n".join([header] + body)
+            if date_f:
+                final_output += "Has been date filtered: {}".format(date_f)
         else:
             final_output = "No matching entries found"
-        log("Action show-sel completed:\n{}".format(final_output))
-        action_set({"message": final_output})
+        hookenv.log("Action show-sel completed:\n{}".format(final_output))
+        hookenv.action_set({"message": final_output})
     except subprocess.CalledProcessError as e:
-        action_fail("Action failed with {}".format(e))
+        hookenv.action_fail("Action failed with {}".format(e))
+
+
+def ack_sel():
+    """Set the acknowledgement date filter."""
+    clear_flag("hw-health.configured")
+
+    date_filter = hookenv.action_get("date")
+    if not date_filter:
+        unitdata.kv().set("date_filter", datetime.now().strftime("%m/%d/%Y"))
+        unitdata.kv().flush()
+        message = "Filtering from now: " + datetime.now().strftime("%m/%d/%Y")
+        hookenv.action_set({"message": message})
+        return
+
+    try:
+        date_time_obj = datetime.strptime(date_filter, "%m/%d/%Y")
+        unitdata.kv().set("date_filter", date_time_obj.strftime("%m/%d/%Y"))
+        unitdata.kv().flush()
+        message = "Filtering from now: " + date_time_obj.strftime("%m/%d/%Y")
+        hookenv.action_set({"message": message})
+    except ValueError:
+        hookenv.action_fail("Action failed with date={} is invalid".format(date_filter))
 
 
 # A dictionary of all the defined actions to callables (which take
@@ -72,16 +119,27 @@ def show_sel():
 ACTIONS = {
     "clear-sel": clear_sel,
     "show-sel": show_sel,
+    "ack-sel": ack_sel,
 }
 
 
 def main(args):
     """Define main subroutine."""
     action_name = os.path.basename(args[0])
+
     try:
         ACTIONS[action_name]()
     except Exception as e:
-        action_fail(str(e))
+        hookenv.action_fail(str(e))
+    else:
+        # try running handlers based on new state
+        unitdata.kv().flush()
+        try:
+            charms.reactive.main()
+        except Exception:
+            exc = format_exc()
+            hookenv.log(exc, hookenv.ERROR)
+            hookenv.action_fail(exc.splitlines()[-1])
 
 
 if __name__ == "__main__":
diff --git a/src/lib/hwhealth/tools.py b/src/lib/hwhealth/tools.py
index a3a92fa..ecc42ef 100644
--- a/src/lib/hwhealth/tools.py
+++ b/src/lib/hwhealth/tools.py
@@ -14,7 +14,7 @@ from zipfile import BadZipFile, ZipFile
 
 from charmhelpers import fetch
 from charmhelpers.contrib.charmsupport import nrpe
-from charmhelpers.core import hookenv
+from charmhelpers.core import hookenv, unitdata
 from charmhelpers.core.host import lsb_release, write_file
 from charmhelpers.core.templating import render
 
@@ -233,6 +233,14 @@ class Tool:
             minutes=",".join(minutes_offsets), user=cron_user, cmd=" ".join(cmdline)
         )
 
+        db = unitdata.kv()
+        if db.get("date_filter") and self._cron_script == "cron_ipmi_sensors.py":
+            hookenv.log("Added date filter to ipmi sensors cmdline", hookenv.DEBUG)
+            cronjob_line = cronjob_line.rstrip()
+            cronjob_line += (
+                ' --seloptions " --date-range ' + db.get("date_filter") + '-now"'
+            )
+
         crond_file = os.path.join(self.CROND_DIR, "hwhealth_{}".format(self._shortname))
         with open(crond_file, "w") as crond_fd:
             crond_fd.write(cronjob_line)