initial code repo
[stor4nfv.git] / src / ceph / src / pybind / mgr / prometheus / module.py
diff --git a/src/ceph/src/pybind/mgr/prometheus/module.py b/src/ceph/src/pybind/mgr/prometheus/module.py
new file mode 100644 (file)
index 0000000..842517f
--- /dev/null
@@ -0,0 +1,405 @@
+import cherrypy
+import json
+import errno
+import math
+import os
+from collections import OrderedDict
+from mgr_module import MgrModule
+
+# Defaults for the Prometheus HTTP server.  Can also set in config-key
+# see https://github.com/prometheus/prometheus/wiki/Default-port-allocations
+# for Prometheus exporter port registry
+
+DEFAULT_ADDR = '::'
+DEFAULT_PORT = 9283
+
+
+# cherrypy likes to sys.exit on error.  don't let it take us down too!
+def os_exit_noop(*args, **kwargs):
+    pass
+
+
+os._exit = os_exit_noop
+
+
+# to access things in class Module from subclass Root.  Because
+# it's a dict, the writer doesn't need to declare 'global' for access
+
+_global_instance = {'plugin': None}
+
+
+def global_instance():
+    assert _global_instance['plugin'] is not None
+    return _global_instance['plugin']
+
+
+def health_status_to_number(status):
+
+    if status == 'HEALTH_OK':
+        return 0
+    elif status == 'HEALTH_WARN':
+        return 1
+    elif status == 'HEALTH_ERR':
+        return 2
+
+PG_STATES = ['creating', 'active', 'clean', 'down', 'scrubbing', 'degraded',
+        'inconsistent', 'peering', 'repair', 'recovering', 'forced-recovery',
+        'backfill', 'forced-backfill', 'wait-backfill', 'backfill-toofull',
+        'incomplete', 'stale', 'remapped', 'undersized', 'peered']
+
+DF_CLUSTER = ['total_bytes', 'total_used_bytes', 'total_objects']
+
+DF_POOL = ['max_avail', 'bytes_used', 'raw_bytes_used', 'objects', 'dirty',
+           'quota_bytes', 'quota_objects', 'rd', 'rd_bytes', 'wr', 'wr_bytes']
+
+OSD_METADATA = ('cluster_addr', 'device_class', 'id', 'public_addr')
+
+OSD_STATUS = ['weight', 'up', 'in']
+
+POOL_METADATA = ('pool_id', 'name')
+
+DISK_OCCUPATION = ('instance', 'device', 'ceph_daemon')
+
+
+class Metric(object):
+    def __init__(self, mtype, name, desc, labels=None):
+        self.mtype = mtype
+        self.name = name
+        self.desc = desc
+        self.labelnames = labels    # tuple if present
+        self.value = dict()         # indexed by label values
+
+    def set(self, value, labelvalues=None):
+        # labelvalues must be a tuple
+        labelvalues = labelvalues or ('',)
+        self.value[labelvalues] = value
+
+    def str_expfmt(self):
+
+        def promethize(path):
+            ''' replace illegal metric name characters '''
+            result = path.replace('.', '_').replace('+', '_plus').replace('::', '_')
+
+            # Hyphens usually turn into underscores, unless they are
+            # trailing
+            if result.endswith("-"):
+                result = result[0:-1] + "_minus"
+            else:
+                result = result.replace("-", "_")
+
+            return "ceph_{0}".format(result)
+
+        def floatstr(value):
+            ''' represent as Go-compatible float '''
+            if value == float('inf'):
+                return '+Inf'
+            if value == float('-inf'):
+                return '-Inf'
+            if math.isnan(value):
+                return 'NaN'
+            return repr(float(value))
+
+        name = promethize(self.name)
+        expfmt = '''
+# HELP {name} {desc}
+# TYPE {name} {mtype}'''.format(
+            name=name,
+            desc=self.desc,
+            mtype=self.mtype,
+        )
+
+        for labelvalues, value in self.value.items():
+            if self.labelnames:
+                labels = zip(self.labelnames, labelvalues)
+                labels = ','.join('%s="%s"' % (k, v) for k, v in labels)
+            else:
+                labels = ''
+            if labels:
+                fmtstr = '\n{name}{{{labels}}} {value}'
+            else:
+                fmtstr = '\n{name} {value}'
+            expfmt += fmtstr.format(
+                name=name,
+                labels=labels,
+                value=floatstr(value),
+            )
+        return expfmt
+
+
+class Module(MgrModule):
+    COMMANDS = [
+        {
+            "cmd": "prometheus self-test",
+            "desc": "Run a self test on the prometheus module",
+            "perm": "rw"
+        },
+    ]
+
+    def __init__(self, *args, **kwargs):
+        super(Module, self).__init__(*args, **kwargs)
+        self.notified = False
+        self.serving = False
+        self.metrics = self._setup_static_metrics()
+        self.schema = OrderedDict()
+        _global_instance['plugin'] = self
+
+    def _stattype_to_str(self, stattype):
+
+        typeonly = stattype & self.PERFCOUNTER_TYPE_MASK
+        if typeonly == 0:
+            return 'gauge'
+        if typeonly == self.PERFCOUNTER_LONGRUNAVG:
+            # this lie matches the DaemonState decoding: only val, no counts
+            return 'counter'
+        if typeonly == self.PERFCOUNTER_COUNTER:
+            return 'counter'
+        if typeonly == self.PERFCOUNTER_HISTOGRAM:
+            return 'histogram'
+
+        return ''
+
+    def _setup_static_metrics(self):
+        metrics = {}
+        metrics['health_status'] = Metric(
+            'untyped',
+            'health_status',
+            'Cluster health status'
+        )
+        metrics['mon_quorum_count'] = Metric(
+            'gauge',
+            'mon_quorum_count',
+            'Monitors in quorum'
+        )
+        metrics['osd_metadata'] = Metric(
+            'untyped',
+            'osd_metadata',
+            'OSD Metadata',
+            OSD_METADATA
+        )
+
+        # The reason for having this separate to OSD_METADATA is
+        # so that we can stably use the same tag names that
+        # the Prometheus node_exporter does
+        metrics['disk_occupation'] = Metric(
+            'undef',
+            'disk_occupation',
+            'Associate Ceph daemon with disk used',
+            DISK_OCCUPATION
+        )
+
+        metrics['pool_metadata'] = Metric(
+            'untyped',
+            'pool_metadata',
+            'POOL Metadata',
+            POOL_METADATA
+        )
+        for state in OSD_STATUS:
+            path = 'osd_{}'.format(state)
+            self.log.debug("init: creating {}".format(path))
+            metrics[path] = Metric(
+                'untyped',
+                path,
+                'OSD status {}'.format(state),
+                ('ceph_daemon',)
+            )
+        for state in PG_STATES:
+            path = 'pg_{}'.format(state)
+            self.log.debug("init: creating {}".format(path))
+            metrics[path] = Metric(
+                'gauge',
+                path,
+                'PG {}'.format(state),
+            )
+        for state in DF_CLUSTER:
+            path = 'cluster_{}'.format(state)
+            self.log.debug("init: creating {}".format(path))
+            metrics[path] = Metric(
+                'gauge',
+                path,
+                'DF {}'.format(state),
+            )
+        for state in DF_POOL:
+            path = 'pool_{}'.format(state)
+            self.log.debug("init: creating {}".format(path))
+            metrics[path] = Metric(
+                'gauge',
+                path,
+                'DF pool {}'.format(state),
+                ('pool_id',)
+            )
+
+        return metrics
+
+    def shutdown(self):
+        self.serving = False
+        pass
+
+    def get_health(self):
+        health = json.loads(self.get('health')['json'])
+        self.metrics['health_status'].set(
+            health_status_to_number(health['status'])
+        )
+
+    def get_df(self):
+        # maybe get the to-be-exported metrics from a config?
+        df = self.get('df')
+        for stat in DF_CLUSTER:
+            path = 'cluster_{}'.format(stat)
+            self.metrics[path].set(df['stats'][stat])
+
+        for pool in df['pools']:
+            for stat in DF_POOL:
+                path = 'pool_{}'.format(stat)
+                self.metrics[path].set(pool['stats'][stat], (pool['id'],))
+
+    def get_quorum_status(self):
+        mon_status = json.loads(self.get('mon_status')['json'])
+        self.metrics['mon_quorum_count'].set(len(mon_status['quorum']))
+
+    def get_pg_status(self):
+        # TODO add per pool status?
+        pg_s = self.get('pg_summary')['all']
+        reported_pg_s = [(s,v) for key, v in pg_s.items() for s in
+                         key.split('+')]
+        for state, value in reported_pg_s:
+            path = 'pg_{}'.format(state)
+            self.metrics[path].set(value)
+        reported_states = [s[0] for s in reported_pg_s]
+        for state in PG_STATES:
+            path = 'pg_{}'.format(state)
+            if state not in reported_states:
+                self.metrics[path].set(0)
+
+    def get_metadata_and_osd_status(self):
+        osd_map = self.get('osd_map')
+        osd_devices = self.get('osd_map_crush')['devices']
+        for osd in osd_map['osds']:
+            id_ = osd['osd']
+            p_addr = osd['public_addr'].split(':')[0]
+            c_addr = osd['cluster_addr'].split(':')[0]
+            dev_class = next((osd for osd in osd_devices if osd['id'] == id_))
+            self.metrics['osd_metadata'].set(0, (
+                c_addr,
+                dev_class['class'],
+                id_,
+                p_addr
+            ))
+            for state in OSD_STATUS:
+                status = osd[state]
+                self.metrics['osd_{}'.format(state)].set(
+                    status,
+                    ('osd.{}'.format(id_),))
+
+            osd_metadata = self.get_metadata("osd", str(id_))
+            dev_keys = ("backend_filestore_dev_node", "bluestore_bdev_dev_node")
+            osd_dev_node = None
+            for dev_key in dev_keys:
+                val = osd_metadata.get(dev_key, None)
+                if val and val != "unknown":
+                    osd_dev_node = val
+                    break
+            osd_hostname = osd_metadata.get('hostname', None)
+            if osd_dev_node and osd_hostname:
+                self.log.debug("Got dev for osd {0}: {1}/{2}".format(
+                    id_, osd_hostname, osd_dev_node))
+                self.metrics['disk_occupation'].set(0, (
+                    osd_hostname,
+                    osd_dev_node,
+                    "osd.{0}".format(id_)
+                ))
+            else:
+                self.log.info("Missing dev node metadata for osd {0}, skipping "
+                               "occupation record for this osd".format(id_))
+
+        for pool in osd_map['pools']:
+            id_ = pool['pool']
+            name = pool['pool_name']
+            self.metrics['pool_metadata'].set(0, (id_, name))
+
+    def collect(self):
+        self.get_health()
+        self.get_df()
+        self.get_quorum_status()
+        self.get_metadata_and_osd_status()
+        self.get_pg_status()
+
+        for daemon, counters in self.get_all_perf_counters().iteritems():
+            for path, counter_info in counters.items():
+                stattype = self._stattype_to_str(counter_info['type'])
+                # XXX simplify first effort: no histograms
+                # averages are already collapsed to one value for us
+                if not stattype or stattype == 'histogram':
+                    self.log.debug('ignoring %s, type %s' % (path, stattype))
+                    continue
+
+                if path not in self.metrics:
+                    self.metrics[path] = Metric(
+                        stattype,
+                        path,
+                        counter_info['description'],
+                        ("ceph_daemon",),
+                    )
+
+                self.metrics[path].set(
+                    counter_info['value'],
+                    (daemon,)
+                )
+
+        return self.metrics
+
+    def handle_command(self, cmd):
+        if cmd['prefix'] == 'prometheus self-test':
+            self.collect()
+            return 0, '', 'Self-test OK'
+        else:
+            return (-errno.EINVAL, '',
+                    "Command not found '{0}'".format(cmd['prefix']))
+
+    def serve(self):
+
+        class Root(object):
+
+            # collapse everything to '/'
+            def _cp_dispatch(self, vpath):
+                cherrypy.request.path = ''
+                return self
+
+            def format_metrics(self, metrics):
+                formatted = ''
+                for m in metrics.values():
+                    formatted += m.str_expfmt()
+                return formatted + '\n'
+
+            @cherrypy.expose
+            def index(self):
+                return '''<!DOCTYPE html>
+<html>
+       <head><title>Ceph Exporter</title></head>
+       <body>
+               <h1>Ceph Exporter</h1>
+               <p><a href='/metrics'>Metrics</a></p>
+       </body>
+</html>'''
+
+            @cherrypy.expose
+            def metrics(self):
+                metrics = global_instance().collect()
+                cherrypy.response.headers['Content-Type'] = 'text/plain'
+                if metrics:
+                    return self.format_metrics(metrics)
+
+        server_addr = self.get_localized_config('server_addr', DEFAULT_ADDR)
+        server_port = self.get_localized_config('server_port', DEFAULT_PORT)
+        self.log.info(
+            "server_addr: %s server_port: %s" %
+            (server_addr, server_port)
+        )
+
+        cherrypy.config.update({
+            'server.socket_host': server_addr,
+            'server.socket_port': int(server_port),
+            'engine.autoreload.on': False
+        })
+        cherrypy.tree.mount(Root(), "/")
+        cherrypy.engine.start()
+        cherrypy.engine.block()