1 from bigquery_analytics import BigQueryAnalytics
7 if os.path.exists("/home/smbaker/projects/vicci/plstackapi/planetstack"):
8 sys.path.append("/home/smbaker/projects/vicci/plstackapi/planetstack")
10 sys.path.append("/opt/planetstack")
12 os.environ.setdefault("DJANGO_SETTINGS_MODULE", "planetstack.settings")
14 from django.db import connection
15 from core.models import Slice, Sliver, ServiceClass, Reservation, Tag, Network, User, Node, Image, Deployment, Site, NetworkTemplate, NetworkSlice, Service
20 class PlanetStackAnalytics(BigQueryAnalytics):
21 def __init__(self, tableName="demoevents"):
22 BigQueryAnalytics.__init__(self, tableName)
24 def service_to_sliceNames(self, serviceName):
25 service=Service.objects.get(name=serviceName)
27 slices = service.slices.all()
29 # BUG in data model -- Slice.service has related name 'service' and
30 # it should be 'slices'
31 slices = service.service.all()
33 return [slice.name for slice in slices]
35 def compose_query(self, slice=None, site=None, node=None, service=None, timeField="MinuteTime", avg=[], sum=[], count=[], computed=[], groupBy=["MinuteTime"], orderBy=["MinuteTime"], tableName="demoevents"):
36 tablePart = "%s.%s@-3600000--1" % ("vicci", tableName)
41 if (timeField=="MinuteTime"):
42 fields.append("INTEGER(TIMESTAMP_TO_SEC(time)/60)*60 as MinuteTime")
43 elif (timeField=="HourTime"):
44 fields.append("INTEGER(TIMESTAMP_TO_SEC(time)/60/60)*60*60 as HourTime")
45 elif (timeField=="DayTime"):
46 fields.append("INTEGER(TIMESTAMP_TO_SEC(time)/60/60/24)*60*60*24 as DayTime")
49 fields.append("AVG(%s) as avg_%s" % (fieldName, fieldName.replace("%","")))
50 fieldNames.append("avg_%s" % fieldName.replace("%",""))
53 fields.append("SUM(%s) as sum_%s" % (fieldName, fieldName.replace("%","")))
54 fieldNames.append("sum_%s" % fieldName.replace("%",""))
56 for fieldName in count:
57 fields.append("COUNT(distinct %s) as count_%s" % (fieldName, fieldName.replace("%","")))
58 fieldNames.append("count_%s" % fieldName.replace("%",""))
60 for fieldName in computed:
62 parts = fieldName.split("/")
63 computedFieldName = "computed_" + parts[0].replace("%","")+"_div_"+parts[1].replace("%","")
66 parts = computed.split("*")
67 computedFieldName = "computed_" + parts[0].replace("%","")+"_mult_"+parts[1].replace("%","")
68 fields.append("SUM(%s)%sSUM(%s) as %s" % (parts[0], operator, parts[1], computedFieldName))
69 fieldNames.append(computedFieldName)
71 for fieldName in groupBy:
72 if (fieldName not in ["MinuteTime", "HourTime", "DayTime"]):
73 fields.append(fieldName)
74 fieldNames.append(fieldName)
76 fields = ", ".join(fields)
81 where.append("%%slice='%s'" % slice)
83 where.append("%%site='%s'" % site)
85 where.append("%%hostname='%s'" % node)
87 sliceNames = self.service_to_sliceNames(service)
89 where.append("(" + " OR ".join(["%%slice='%s'" % sliceName for sliceName in sliceNames]) +")")
92 where = " WHERE " + " AND ".join(where)
97 groupBySub = " GROUP BY " + ",".join(groupBy + ["%hostname"])
98 groupBy = " GROUP BY " + ",".join(groupBy)
100 groupBySub = " GROUP BY %hostname"
104 orderBy = " ORDER BY " + ",".join(orderBy)
109 subQuery = "SELECT %%hostname, %s FROM [%s]" % (fields, tablePart)
111 subQuery = subQuery + where
112 subQuery = subQuery + groupBySub
113 #subQuery = subQuery + " GROUP BY %s,%%hostname" % timeField
116 for fieldName in fieldNames:
117 if fieldName.startswith("avg"):
118 sumFields.append("AVG(%s) as avg_%s"%(fieldName,fieldName))
119 sumFields.append("MAX(%s) as max_%s"%(fieldName,fieldName))
120 elif (fieldName.startswith("count")) or (fieldName.startswith("sum")) or (fieldName.startswith("computed")):
121 sumFields.append("SUM(%s) as sum_%s"%(fieldName,fieldName))
123 sumFields.append(fieldName)
125 sumFields = ",".join(sumFields)
127 query = "SELECT %s, %s FROM (%s)" % (timeField, sumFields, subQuery)
129 query = query + groupBy
131 query = query + orderBy
133 query = "SELECT %s FROM [%s]" % (fields, tablePart)
135 query = query + " " + where
137 query = query + groupBy
139 query = query + orderBy
143 def get_list_from_req(self, req, name, default=[]):
144 value = req.GET.get(name, None)
147 return value.split(",")
149 def format_result(self, format, result, query):
150 if (format == "json_dicts"):
151 result = {"query": query, "rows": result}
152 return ("application/javascript", json.dumps(result))
154 elif (format == "json_arrays"):
158 for key in sorted(row.keys()):
159 new_row.append(row[key])
160 new_result.append(new_row)
161 new_result = {"query": query, "rows": new_result}
162 return ("application/javascript", json.dumps(new_result))
164 elif (format == "html_table"):
168 for key in sorted(row.keys()):
169 new_row.append("<TD>%s</TD>" % str(row[key]))
170 new_rows.append("<TR>%s</TR>" % "".join(new_row))
172 new_result = "<TABLE>%s</TABLE>" % "\n".join(new_rows)
174 return ("text/html", new_result)
176 elif (format == "json_hpcdash"):
179 new_row = {"lat": float(row.get("lat", 0)),
180 "long": float(row.get("long", 0)),
182 "numNodes": int(row.get("numNodes",0)),
183 "numHPCSlivers": int(row.get("sum_count_hostname", 0)),
184 "siteUrl": row.get("url", ""),
185 "hot": float(row.get("hotness", 0.0)),
186 "load": int(float(row.get("max_avg_cpu", 0)))}
187 new_rows[row["site"]] = new_row
188 return ("application/javascript", json.dumps(new_rows))
190 def only_largest(self, rows, fieldName):
191 """ Given a fieldName, only return the set of rows that had the
192 maximum value of that fieldName.
194 maxVal = max( [int(row[fieldName]) for row in rows] )
195 new_rows = [row for row in rows if int(row[fieldName])==maxVal]
198 def merge_datamodel_sites(self, rows):
199 """ For a query that included "site" in its groupby, merge in the
200 opencloud site information.
203 sitename = row["site"]
205 model_site = Site.objects.get(name=sitename)
207 # we didn't find it in the data model
210 row["lat"] = float(model_site.location.latitude)
211 row["long"] = float(model_site.location.longitude)
212 row["url"] = model_site.site_url
213 row["numNodes"] = model_site.nodes.count()
215 if "max_avg_cpu" in row:
216 cpu=float(row["max_avg_cpu"])/100.0
217 row["hotness"] = max(0.0, ((cpu*RED_LOAD) - BLUE_LOAD)/(RED_LOAD-BLUE_LOAD))
219 def process_request(self, req):
222 tqx = req.GET.get("reqId", None)
224 slice = req.GET.get("slice", None)
225 site = req.GET.get("site", None)
226 node = req.GET.get("node", None)
227 service = req.GET.get("service", None)
229 format = req.GET.get("format", "json_dicts")
231 timeField = req.GET.get("timeField", "MinuteTime")
232 avg = self.get_list_from_req(req, "avg")
233 sum = self.get_list_from_req(req, "sum")
234 count = self.get_list_from_req(req, "count")
235 computed = self.get_list_from_req(req, "computed")
236 groupBy = self.get_list_from_req(req, "groupBy", ["MinuteTime"])
237 orderBy = self.get_list_from_req(req, "orderBy", ["MinuteTime"])
239 maxRows = req.GET.get("maxRows", None)
240 onlyLargest = req.GET.get("onlyLargest", None)
241 mergeDataModelSites = req.GET.get("mergeDataModelSites", None)
243 q = self.compose_query(slice, site, node, service, timeField, avg, sum, count, computed, groupBy, orderBy)
248 result = self.run_query_raw(q)
249 result["reqId"] = 0 # XXX FIXME
250 return ("application/javascript", json.dumps(result))
252 result = self.run_query(q)
255 result = self.only_largest(result, onlyLargest)
257 if mergeDataModelSites:
258 self.merge_datamodel_sites(result)
261 result = result[-int(maxRows):]
263 return self.format_result(format, result, q)
266 def DoPlanetStackAnalytics(request):
267 bq = PlanetStackAnalytics()
268 result = bq.process_request(request)
273 bq = PlanetStackAnalytics()
276 q=bq.compose_query(avg=["%cpu"], count=["%hostname"], slice="HyperCache")
278 bq.dump_table(bq.run_query(q))
280 q=bq.compose_query(computed=["%bytes_sent/%elapsed"])
283 bq.dump_table(bq.run_query(q))
285 q=bq.compose_query(timeField="HourTime", avg=["%cpu"], count=["%hostname"], computed=["%bytes_sent/%elapsed"], groupBy=["HourTime"], orderBy=["HourTime"])
288 bq.dump_table(bq.run_query(q))
291 q=bq.compose_query(avg=["%cpu"], count=["%hostname"], computed=["%bytes_sent/%elapsed"], service="HPC Service", groupBy=["MinuteTime","%site"])
294 result=bq.run_query(q)
295 result = bq.only_largest(result, "MinuteTime")
296 bq.merge_datamodel_sites(result)
297 #bq.dump_table(result)
298 print bq.format_result("json_hpcdash", result, q)
300 if __name__ == "__main__":