Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Backport 1.5: Disable usage metrics on performance standby nodes. (#9966) #9967

Merged
merged 2 commits into from
Sep 22, 2020
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion vault/core.go
Original file line number Diff line number Diff line change
Expand Up @@ -1971,7 +1971,8 @@ func (s standardUnsealStrategy) unseal(ctx context.Context, logger log.Logger, c
return nil
}

// postUnseal is invoked on the active node after the barrier is unsealed, but before
// postUnseal is invoked on the active node, and performance standby nodes,
// after the barrier is unsealed, but before
// allowing any user operations. This allows us to setup any state that
// requires the Vault to be unsealed such as mount tables, logical backends,
// credential stores, etc.
Expand Down
18 changes: 15 additions & 3 deletions vault/core_metrics.go
Original file line number Diff line number Diff line change
Expand Up @@ -27,10 +27,13 @@ func (c *Core) metricsLoop(stopCh chan struct{}) {
select {
case <-emitTimer:
c.metricsMutex.Lock()
if c.expiration != nil {

// Emit on active node only
if c.expiration != nil && !c.perfStandby {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggest c.PerfStandby() instead for locking.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think that some care is needed because c.stateLock can be held while acquiring c.metricsLock, so we shouldn't acquire them in the reverse order here.

c.expiration.emitMetrics()
}
// Refresh the sealed gauge

// Refresh the sealed gauge, on all nodes
if c.Sealed() {
c.metricSink.SetGaugeWithLabels([]string{"core", "unsealed"}, 0, nil)
} else {
Expand All @@ -54,6 +57,11 @@ func (c *Core) metricsLoop(stopCh chan struct{}) {
}
c.stateLock.RUnlock()
case <-identityCountTimer:
// Only emit on active node
if c.perfStandby {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ditto.

break
}

// TODO: this can be replaced by the identity gauge counter; we need to
// sum across all namespaces.
go func() {
Expand Down Expand Up @@ -123,6 +131,9 @@ func (c *Core) emitMetrics(stopCh chan struct{}) {
// The gauge collection processes are started and stopped here
// because there's more than one TokenManager created during startup,
// but we only want one set of gauges.
//
// Both active nodes and performance standby nodes call emitMetrics
// so we have to handle both.

metricsInit := []struct {
MetricName []string
Expand Down Expand Up @@ -174,9 +185,10 @@ func (c *Core) emitMetrics(stopCh chan struct{}) {
},
}

// Disable collection if configured, or if we're a performance standby.
if c.MetricSink().GaugeInterval == time.Duration(0) {
c.logger.Info("usage gauge collection is disabled")
} else {
} else if !c.perfStandby {
for _, init := range metricsInit {
if init.DisableEnvVar != "" {
if os.Getenv(init.DisableEnvVar) != "" {
Expand Down