1 from bigquery_analytics import BigQueryAnalytics
11 if os.path.exists("/home/smbaker/projects/vicci/plstackapi/planetstack"):
12 sys.path.append("/home/smbaker/projects/vicci/plstackapi/planetstack")
14 sys.path.append("/opt/planetstack")
16 os.environ.setdefault("DJANGO_SETTINGS_MODULE", "planetstack.settings")
18 from django.db import connection
19 from core.models import Slice, Sliver, ServiceClass, Reservation, Tag, Network, User, Node, Image, Deployment, Site, NetworkTemplate, NetworkSlice, Service
24 glo_cached_queries = {}
26 class PlanetStackAnalytics(BigQueryAnalytics):
27 def __init__(self, tableName="demoevents"):
28 BigQueryAnalytics.__init__(self, tableName)
30 def service_to_sliceNames(self, serviceName):
31 service=Service.objects.get(name=serviceName)
33 slices = service.slices.all()
35 # BUG in data model -- Slice.service has related name 'service' and
36 # it should be 'slices'
37 slices = service.service.all()
39 return [slice.name for slice in slices]
41 def compose_query(self, slice=None, site=None, node=None, service=None, timeBucket="60", avg=[], sum=[], count=[], computed=[], val=[], groupBy=["Time"], orderBy=["Time"], tableName="demoevents", latest=False):
42 tablePart = "[%s.%s@-3600000--1]" % ("vicci", tableName)
46 srcFieldNames = ["time"]
48 fields.append("SEC_TO_TIMESTAMP(INTEGER(TIMESTAMP_TO_SEC(time)/%s)*%s) as Time" % (str(timeBucket),str(timeBucket)))
49 #fields.append("INTEGER(TIMESTAMP_TO_SEC(time)/%s)*%s as Time" % (str(timeBucket),str(timeBucket)))
52 fields.append("AVG(%s) as avg_%s" % (fieldName, fieldName.replace("%","")))
53 fieldNames.append("avg_%s" % fieldName.replace("%",""))
54 srcFieldNames.append(fieldName)
57 fields.append("SUM(%s) as sum_%s" % (fieldName, fieldName.replace("%","")))
58 fieldNames.append("sum_%s" % fieldName.replace("%",""))
59 srcFieldNames.append(fieldName)
61 for fieldName in count:
62 fields.append("COUNT(distinct %s) as count_%s" % (fieldName, fieldName.replace("%","")))
63 fieldNames.append("count_%s" % fieldName.replace("%",""))
64 srcFieldNames.append(fieldName)
67 fields.append(fieldName)
68 fieldNames.append(fieldName)
69 srcFieldNames.append(fieldName)
71 for fieldName in computed:
73 parts = fieldName.split("/")
74 computedFieldName = "computed_" + parts[0].replace("%","")+"_div_"+parts[1].replace("%","")
77 parts = computed.split("*")
78 computedFieldName = "computed_" + parts[0].replace("%","")+"_mult_"+parts[1].replace("%","")
79 fields.append("SUM(%s)%sSUM(%s) as %s" % (parts[0], operator, parts[1], computedFieldName))
80 fieldNames.append(computedFieldName)
81 srcFieldNames.append(parts[0])
82 srcFieldNames.append(parts[1])
84 for fieldName in groupBy:
85 if (fieldName not in ["Time"]):
86 fields.append(fieldName)
87 fieldNames.append(fieldName)
88 srcFieldNames.append(fieldName)
90 fields = ", ".join(fields)
95 where.append("%%slice='%s'" % slice)
97 where.append("%%site='%s'" % site)
99 where.append("%%hostname='%s'" % node)
101 sliceNames = self.service_to_sliceNames(service)
103 where.append("(" + " OR ".join(["%%slice='%s'" % sliceName for sliceName in sliceNames]) +")")
106 where = " WHERE " + " AND ".join(where)
111 groupBySub = " GROUP BY " + ",".join(groupBy + ["%hostname"])
112 groupBy = " GROUP BY " + ",".join(groupBy)
114 groupBySub = " GROUP BY %hostname"
118 orderBy = " ORDER BY " + ",".join(orderBy)
123 latestFields = ["table1.%s as %s" % (x,x) for x in srcFieldNames]
124 latestFields = ", ".join(latestFields)
125 tablePart = """(SELECT %s FROM %s AS table1
127 (SELECT %%hostname, event, max(time) as maxtime from %s GROUP BY %%hostname, event) AS latest
129 table1.%%hostname = latest.%%hostname AND table1.event = latest.event AND table1.time = latest.maxtime)""" % (latestFields, tablePart, tablePart)
132 subQuery = "SELECT %%hostname, %s FROM %s" % (fields, tablePart)
134 subQuery = subQuery + where
135 subQuery = subQuery + groupBySub
138 for fieldName in fieldNames:
139 if fieldName.startswith("avg"):
140 sumFields.append("AVG(%s) as avg_%s"%(fieldName,fieldName))
141 sumFields.append("MAX(%s) as max_%s"%(fieldName,fieldName))
142 elif (fieldName.startswith("count")) or (fieldName.startswith("sum")) or (fieldName.startswith("computed")):
143 sumFields.append("SUM(%s) as sum_%s"%(fieldName,fieldName))
145 sumFields.append(fieldName)
147 sumFields = ",".join(sumFields)
149 query = "SELECT %s, %s FROM (%s)" % ("Time", sumFields, subQuery)
151 query = query + groupBy
153 query = query + orderBy
155 query = "SELECT %s FROM %s" % (fields, tablePart)
157 query = query + " " + where
159 query = query + groupBy
161 query = query + orderBy
165 def get_list_from_req(self, req, name, default=[]):
166 value = req.GET.get(name, None)
169 value=value.replace("@","%")
170 return value.split(",")
172 def format_result(self, format, result, query, dataSourceUrl):
173 if (format == "json_dicts"):
174 result = {"query": query, "rows": result, "dataSourceUrl": dataSourceUrl}
175 return ("application/javascript", json.dumps(result))
177 elif (format == "json_arrays"):
181 for key in sorted(row.keys()):
182 new_row.append(row[key])
183 new_result.append(new_row)
184 new_result = {"query": query, "rows": new_result}
185 return ("application/javascript", json.dumps(new_result))
187 elif (format == "html_table"):
191 for key in sorted(row.keys()):
192 new_row.append("<TD>%s</TD>" % str(row[key]))
193 new_rows.append("<TR>%s</TR>" % "".join(new_row))
195 new_result = "<TABLE>%s</TABLE>" % "\n".join(new_rows)
197 return ("text/html", new_result)
199 def merge_datamodel_sites(self, rows):
200 """ For a query that included "site" in its groupby, merge in the
201 opencloud site information.
204 sitename = row["site"]
206 model_site = Site.objects.get(name=sitename)
208 # we didn't find it in the data model
211 row["lat"] = float(model_site.location.latitude)
212 row["long"] = float(model_site.location.longitude)
213 row["url"] = model_site.site_url
214 row["numNodes"] = model_site.nodes.count()
216 max_cpu = row.get("max_avg_cpu", row.get("max_cpu",0))
217 cpu=float(max_cpu)/100.0
218 row["hotness"] = max(0.0, ((cpu*RED_LOAD) - BLUE_LOAD)/(RED_LOAD-BLUE_LOAD))
220 def compose_latest_query(self, fieldNames=None, groupByFields=["%hostname", "event"]):
221 """ Compose a query that returns the 'most recent' row for each (hostname, event)
226 fieldNames = ["%hostname", "%bytes_sent", "time", "event", "%site", "%elapsed", "%slice", "%cpu"]
228 fields = ["table1.%s AS %s" % (x,x) for x in fieldNames]
229 fields = ", ".join(fields)
231 tableDesc = "%s.%s" % (self.projectName, self.tableName)
233 groupByOn = ["table1.time = latest.maxtime"]
234 for field in groupByFields:
235 groupByOn.append("table1.%s = latest.%s" % (field, field))
237 groupByOn = " AND ".join(groupByOn)
238 groupByFields = ", ".join(groupByFields)
240 base_query = "SELECT %s FROM [%s@-3600000--1] AS table1 JOIN (SELECT %s, max(time) as maxtime from [%s@-3600000--1] GROUP BY %s) AS latest ON %s" % \
241 (fields, tableDesc, groupByFields, tableDesc, groupByFields, groupByOn)
245 def get_cached_query_results(self, q):
246 global glo_cached_queries
248 if q in glo_cached_queries:
249 if (time.time() - glo_cached_queries[q]["time"]) <= 60:
250 print "using cached query"
251 return glo_cached_queries[q]["rows"]
253 print "refreshing cached query"
254 result = self.run_query(q)
255 glo_cached_queries[q] = {"time": time.time(), "rows": result}
259 def process_request(self, req):
262 tqx = req.GET.get("tqx", None)
264 slice = req.GET.get("slice", None)
265 site = req.GET.get("site", None)
266 node = req.GET.get("node", None)
267 service = req.GET.get("service", None)
269 format = req.GET.get("format", "json_dicts")
271 timeField = req.GET.get("timeBucket", "60")
272 avg = self.get_list_from_req(req, "avg")
273 sum = self.get_list_from_req(req, "sum")
274 count = self.get_list_from_req(req, "count")
275 computed = self.get_list_from_req(req, "computed")
276 groupBy = self.get_list_from_req(req, "groupBy", ["Time"])
277 orderBy = self.get_list_from_req(req, "orderBy", ["Time"])
279 maxRows = req.GET.get("maxRows", None)
280 mergeDataModelSites = req.GET.get("mergeDataModelSites", None)
282 cached = req.GET.get("cached", None)
284 q = self.compose_query(slice, site, node, service, timeField, avg, sum, count, computed, [], groupBy, orderBy)
288 dataSourceUrl = "http://" + req.META["SERVER_NAME"] + ":" + req.META["SERVER_PORT"] + req.META["PATH_INFO"] + "?" + req.META["QUERY_STRING"].replace("format=","origFormat=").replace("%","%25") + "&format=charts";
290 if (format=="dataSourceUrl"):
291 result = {"dataSourceUrl": dataSourceUrl}
292 return ("application/javascript", result)
294 elif (format=="raw"):
295 result = self.run_query_raw(q)
296 result["dataSourceUrl"] = dataSourceUrl
298 result = json.dumps(result);
300 return ("application/javascript", result)
302 elif (format=="nodata"):
303 result = {"dataSourceUrl": dataSourceUrl, "query": q}
304 result = json.dumps(result);
305 return {"application/javascript", result}
307 elif (format=="charts"):
308 bq_result = self.run_query_raw(q)
310 # cloudscrutiny code is probably better!
312 table["cols"] = self.schema_to_cols(bq_result["schema"])
314 for row in bq_result["rows"]:
316 for (colnum,col) in enumerate(row["f"]):
318 dt = datetime.datetime.fromtimestamp(float(col["v"]))
319 rowcols.append({"v": 'new Date("%s")' % dt.isoformat()})
322 rowcols.append({"v": float(col["v"])})
324 rowcols.append({"v": col["v"]})
325 rows.append({"c": rowcols})
329 reqId = tqx.strip("reqId:")
333 result = {"status": "okColumnChart", "reqId": reqId, "table": table, "version": "0.6"}
335 result = "google.visualization.Query.setResponse(" + json.dumps(result) + ");"
337 def unquote_it(x): return x.group()[1:-1].replace('\\"', '"')
339 p = re.compile(r'"new Date\(\\"[^"]*\\"\)"')
340 result=p.sub(unquote_it, result)
342 return ("application/javascript", result)
346 results = self.get_cached_query_results(self.compose_latest_query())
350 filter["slice"] = slice
352 filter["site"] = site
354 filter["hostname"] = node
356 result = self.postprocess_results(results, filter=filter, sum=sum, count=count, avg=avg, computed=computed, maxDeltaTime=120, groupBy=["doesnotexist"])
358 result = self.run_query(q)
361 result = result[-int(maxRows):]
363 if mergeDataModelSites:
364 self.merge_datamodel_sites(result)
366 return self.format_result(format, result, q, dataSourceUrl)
368 def DoPlanetStackAnalytics(request):
369 bq = PlanetStackAnalytics()
370 result = bq.process_request(request)
375 bq = PlanetStackAnalytics()
377 q = bq.compose_latest_query()
378 results = bq.run_query(q)
380 results = bq.postprocess_results(results,
381 #filter={"site": "Princeton"},
383 computed=["bytes_sent/elapsed"],
384 sum=["bytes_sent", "computed_bytes_sent_div_elapsed"], avg=["cpu"],
387 bq.dump_table(results)
389 q=bq.compose_query(sum=["%bytes_sent"], avg=["%cpu"], latest=True, groupBy=["Time", "%site"])
391 bq.dump_table(bq.run_query(q))
395 q=bq.compose_query(avg=["%cpu","%bandwidth"], count=["%hostname"], slice="HyperCache")
397 bq.dump_table(bq.run_query(q))
399 q=bq.compose_query(computed=["%bytes_sent/%elapsed"])
402 bq.dump_table(bq.run_query(q))
404 q=bq.compose_query(timeBucket=60*60, avg=["%cpu"], count=["%hostname"], computed=["%bytes_sent/%elapsed"])
407 bq.dump_table(bq.run_query(q))
409 if __name__ == "__main__":