17706: when a container request has no container associated (e.g. when

[arvados.git] / lib / costanalyzer / costanalyzer.go

diff --git a/lib/costanalyzer/costanalyzer.go b/lib/costanalyzer/costanalyzer.go
index e8dd186050fb200ca575a480ff6d15db1d88adf0..71cb59093ab2b7c9ba6e7ffcbe71d4cab513acd0 100644 (file)
--- a/lib/costanalyzer/costanalyzer.go
+++ b/lib/costanalyzer/costanalyzer.go
@@ -35,6 +35,7 @@ type nodeInfo struct {
        // Modern
        ProviderType string
        Price        float64
+       Preemptible  bool
 }
 
 type arrayFlags []string
@@ -56,7 +57,7 @@ func parseFlags(prog string, args []string, loader *config.Loader, logger *logru
        flags.Usage = func() {
                fmt.Fprintf(flags.Output(), `
 Usage:
-  %s [options ...] <uuid> ...
+  %s [options ...] uuid [uuid ...]
 
        This program analyzes the cost of Arvados container requests. For each uuid
        supplied, it creates a CSV report that lists all the containers used to
@@ -77,19 +78,26 @@ Usage:
        container was reused between several container requests, its cost will only
        be counted once.
 
-       To get the node costs, the progam queries the Arvados API for current cost
-       data for each node type used. This means that the reported cost always
-       reflects the cost data as currently defined in the Arvados API configuration
-       file.
-
        Caveats:
-       - the Arvados API configuration cost data may be out of sync with the cloud
-       provider.
-       - when generating reports for older container requests, the cost data in the
-       Arvados API configuration file may have changed since the container request
-       was fulfilled. This program uses the cost data stored at the time of the
+
+       - This program uses the cost data from config.yml at the time of the
        execution of the container, stored in the 'node.json' file in its log
-       collection.
+       collection. If the cost data was not correctly configured at the time the
+       container was executed, the output from this program will be incorrect.
+
+       - If a container was run on a preemptible ("spot") instance, the cost data
+       reported by this program may be wildly inaccurate, because it does not have
+       access to the spot pricing in effect for the node then the container ran. The
+       UUID report file that is generated when the '-output' option is specified has
+       a column that indicates the preemptible state of the instance that ran the
+       container.
+
+       - This program does not take into account overhead costs like the time spent
+       starting and stopping compute nodes that run containers, the cost of the
+       permanent cloud nodes that provide the Arvados services, the cost of data
+       stored in Arvados, etc.
+
+       - When provided with a project uuid, subprojects will not be considered.
 
        In order to get the data for the uuids supplied, the ARVADOS_API_HOST and
        ARVADOS_API_TOKEN environment variables must be set.
@@ -181,7 +189,7 @@ func addContainerLine(logger *logrus.Logger, node nodeInfo, cr arvados.Container
                size = node.ProviderType
        }
        cost = delta.Seconds() / 3600 * price
-       csv += size + "," + strconv.FormatFloat(price, 'f', 8, 64) + "," + strconv.FormatFloat(cost, 'f', 8, 64) + "\n"
+       csv += size + "," + fmt.Sprintf("%+v", node.Preemptible) + "," + strconv.FormatFloat(price, 'f', 8, 64) + "," + strconv.FormatFloat(cost, 'f', 8, 64) + "\n"
        return
 }
 
@@ -369,7 +377,7 @@ func generateCrCsv(logger *logrus.Logger, uuid string, arv *arvadosclient.Arvado
 
        cost = make(map[string]float64)
 
-       csv := "CR UUID,CR name,Container UUID,State,Started At,Finished At,Duration in seconds,Compute node type,Hourly node cost,Total cost\n"
+       csv := "CR UUID,CR name,Container UUID,State,Started At,Finished At,Duration in seconds,Compute node type,Preemptible,Hourly node cost,Total cost\n"
        var tmpCsv string
        var tmpTotalCost float64
        var totalCost float64
@@ -398,6 +406,11 @@ func generateCrCsv(logger *logrus.Logger, uuid string, arv *arvadosclient.Arvado
        if err != nil {
                return nil, fmt.Errorf("error loading cr object %s: %s", uuid, err)
        }
+       if len(cr.ContainerUUID) == 0 {
+               // Nothing to do! E.g. a CR in 'Uncommitted' state.
+               logger.Infof("No container associated with container request %s, skipping\n", crUUID)
+               return nil, nil
+       }
        var container arvados.Container
        err = loadObject(logger, ac, crUUID, cr.ContainerUUID, cache, &container)
        if err != nil {