// clusteringHandler handles doing the actual k-means clustering.
//
// The return format is JSON of the form:
//
// {
// "Clusters": [
// {
// "Keys": [
// "x86:GeForce320M:MacMini4.1:Mac10.8:GM_varied_text_clipped_no_lcd_640_480:8888",...],
// "ParamSummaries": [
// [{"Value": "Win8", "Weight": 15}, {"Value": "Android", "Weight": 14}, ...]
// ],
// "StepFit": {
// "LeastSquares":0.0006582442047814354,
// "TurningPoint":162,
// "StepSize":0.023272272692293046,
// "Regression": 35.3
// }
// Traces: [[[0, -0.00007967326606768456], [1, 0.011877665949459049], [2, 0.012158129176717419],...]]
// },
// ...
// ],
// "K": 5,
// "StdDevThreshhold": 0.1
// }
//
// Note that Keys contains all the keys, while Traces only contains traces of
// the N closest cluster members and the centroid.
//
// Takes the following query parameters:
//
// _k - The K to use for k-means clustering.
// _stddev - The standard deviation to use when normalize traces
// during k-means clustering.
// _issue - The Rietveld issue ID with trybot results to include.
//
// Additionally the rest of the query parameters as returned from
// sk.Query.selectionsAsQuery().
func clusteringHandler(w http.ResponseWriter, r *http.Request) {
glog.Infof("Clustering Handler: %q\n", r.URL.Path)
tile := masterTileBuilder.GetTile()
w.Header().Set("Content-Type", "application/json")
// If there are no query parameters just return with an empty set of ClusterSummaries.
if r.FormValue("_k") == "" || r.FormValue("_stddev") == "" {
writeClusterSummaries(clustering.NewClusterSummaries(), w, r)
return
}
k, err := strconv.ParseInt(r.FormValue("_k"), 10, 32)
if err != nil {
util.ReportError(w, r, err, fmt.Sprintf("_k parameter must be an integer %s.", r.FormValue("_k")))
return
}
stddev, err := strconv.ParseFloat(r.FormValue("_stddev"), 64)
if err != nil {
util.ReportError(w, r, err, fmt.Sprintf("_stddev parameter must be a float %s.", r.FormValue("_stddev")))
return
}
issue := r.FormValue("_issue")
var tryResults *types.TryBotResults = nil
if issue != "" {
var err error
tryResults, err = trybot.Get(issue)
if err != nil {
util.ReportError(w, r, err, fmt.Sprintf("Failed to get trybot data for clustering."))
return
}
}
delete(r.Form, "_k")
delete(r.Form, "_stddev")
delete(r.Form, "_issue")
// Create a filter function for traces that match the query parameters and
// optionally tryResults.
filter := func(key string, tr *types.PerfTrace) bool {
if tryResults != nil {
if _, ok := tryResults.Values[key]; !ok {
return false
}
}
return tiling.Matches(tr, r.Form)
}
if issue != "" {
if tile, err = trybot.TileWithTryData(tile, issue); err != nil {
util.ReportError(w, r, err, fmt.Sprintf("Failed to get trybot data for clustering."))
return
}
}
summary, err := clustering.CalculateClusterSummaries(tile, int(k), stddev, filter)
if err != nil {
util.ReportError(w, r, err, "Failed to calculate clusters.")
return
}
writeClusterSummaries(summary, w, r)
}
// singleStep does a single round of alerting.
func singleStep(issueTracker issues.IssueTracker) {
latencyBegin := time.Now()
tile := tileBuilder.GetTile()
summary, err := clustering.CalculateClusterSummaries(tile, CLUSTER_SIZE, CLUSTER_STDDEV, skpOnly)
if err != nil {
glog.Errorf("Alerting: Failed to calculate clusters: %s", err)
return
}
fresh := []*types.ClusterSummary{}
for _, c := range summary.Clusters {
if math.Abs(c.StepFit.Regression) > clustering.INTERESTING_THRESHHOLD {
fresh = append(fresh, c)
}
}
old, err := ListFrom(tile.Commits[0].CommitTime)
if err != nil {
glog.Errorf("Alerting: Failed to get existing clusters: %s", err)
return
}
glog.Infof("Found %d old", len(old))
glog.Infof("Found %d fresh", len(fresh))
updated := CombineClusters(fresh, old)
glog.Infof("Found %d to update", len(updated))
for _, c := range updated {
if c.Status == "" {
c.Status = "New"
}
if err := Write(c); err != nil {
glog.Errorf("Alerting: Failed to write updated cluster: %s", err)
}
}
current, err := ListFrom(tile.Commits[0].CommitTime)
if err != nil {
glog.Errorf("Alerting: Failed to get existing clusters: %s", err)
return
}
count := 0
for _, c := range current {
if c.Status == "New" {
count++
}
if issueTracker != nil {
if err := updateBugs(c, issueTracker); err != nil {
glog.Errorf("Error retrieving bugs: %s", err)
return
}
} else {
glog.Infof("Skipping ClusterSummary.Bugs update because apiKey is missing.")
return
}
}
newClustersGauge.Update(int64(count))
runsCounter.Inc(1)
alertingLatency.UpdateSince(latencyBegin)
}