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.

Sign up for GitHub

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

Jump to bottom

allocrunner: fix health check monitoring for Consul services #16402

Merged
merged 6 commits into from
Mar 10, 2023
Merged

allocrunner: fix health check monitoring for Consul services #16402

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
75bce63
allocrunner: fix health check monitoring for Consul services
lgfa29 Mar 9, 2023
5f8dabf
changelog: add entry for #16402
lgfa29 Mar 9, 2023
933de17
test: fix allocHealthWatcherHook tests
lgfa29 Mar 10, 2023
d209dcd
clarify some comments and emit a warning when task environment is not…
lgfa29 Mar 10, 2023
925d347
client: skip variable interpolation if environment is not available
lgfa29 Mar 10, 2023
2c84b03
test: canonicalze mock.Alloc()
lgfa29 Mar 10, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
3 changes: 3 additions & 0 deletions .changelog/16402.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
```release-note:bug
client: Fixed a bug that prevented allocations with interpolated values in Consul services from being marked as healthy
```
35 changes: 32 additions & 3 deletions client/allochealth/tracker.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,7 @@ import (
"github.com/hashicorp/nomad/client/serviceregistration"
"github.com/hashicorp/nomad/client/serviceregistration/checks/checkstore"
cstructs "github.com/hashicorp/nomad/client/structs"
"github.com/hashicorp/nomad/client/taskenv"
"github.com/hashicorp/nomad/helper"
"github.com/hashicorp/nomad/nomad/structs"
"golang.org/x/exp/maps"
Expand Down Expand Up @@ -96,6 +97,10 @@ type Tracker struct {
// name -> state
taskHealth map[string]*taskHealthState

// taskEnvs maps each task in the allocation to a *taskenv.TaskEnv that is
// used to interpolate runtime variables used in service definitions.
taskEnvs map[string]*taskenv.TaskEnv

// logger is for logging things
logger hclog.Logger
}
Expand All @@ -111,6 +116,7 @@ func NewTracker(
logger hclog.Logger,
alloc *structs.Allocation,
allocUpdates *cstructs.AllocListener,
taskEnvBuilder *taskenv.Builder,
consulClient serviceregistration.Handler,
checkStore checkstore.Shim,
minHealthyTime time.Duration,
Expand All @@ -132,6 +138,12 @@ func NewTracker(
lifecycleTasks: map[string]string{},
}

// Build the map of TaskEnv for each task. Create the group-level TaskEnv
// first because taskEnvBuilder is mutated in every loop and we can't undo
// a call to UpdateTask().
t.taskEnvs = make(map[string]*taskenv.TaskEnv, len(t.tg.Tasks)+1)
t.taskEnvs[""] = taskEnvBuilder.Build()

t.taskHealth = make(map[string]*taskHealthState, len(t.tg.Tasks))
for _, task := range t.tg.Tasks {
t.taskHealth[task.Name] = &taskHealthState{task: task}
Expand All @@ -140,6 +152,8 @@ func NewTracker(
t.lifecycleTasks[task.Name] = task.Lifecycle.Hook
}

t.taskEnvs[task.Name] = taskEnvBuilder.UpdateTask(alloc, task).Build()

c, n := countChecks(task.Services)
t.consulCheckCount += c
t.nomadCheckCount += n
Expand Down Expand Up @@ -503,8 +517,24 @@ OUTER:
// Detect if all the checks are passing
passed := true

// interpolate services to replace runtime variables
consulServices := t.tg.ConsulServices()
interpolatedServices := make([]*structs.Service, len(consulServices))
for i, service := range consulServices {
env := t.taskEnvs[service.TaskName]
if env == nil {
// This is not expected to happen, but guard against a nil
// task environment by using the group environment since it has
// most of the same values.
t.logger.Warn("task environment not found, using group level environment for variable interpolation",
"alloc_id", t.alloc.ID, "task", service.TaskName)
env = t.taskEnvs[""]
Copy link
Member

Choose a reason for hiding this comment

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

we [should] already guard against this in Validate/Canonicalize; instead of trying to make things maybe work in an unexpected situation I think we should just do nothing, e.g.

if env != nil {
 // assign
} else {
  logger.Error("you found a bug!")
}

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Hum...good point. Changed to this in 925d347

}
interpolatedServices[i] = taskenv.InterpolateService(env, service)
}

// scan for missing or unhealthy consul checks
if !evaluateConsulChecks(t.tg, allocReg) {
if !evaluateConsulChecks(interpolatedServices, allocReg) {
t.setCheckHealth(false)
passed = false
}
Expand All @@ -523,11 +553,10 @@ OUTER:
}
}

func evaluateConsulChecks(tg *structs.TaskGroup, registrations *serviceregistration.AllocRegistration) bool {
func evaluateConsulChecks(services []*structs.Service, registrations *serviceregistration.AllocRegistration) bool {
// First, identify any case where a check definition is missing or outdated
// on the Consul side. Note that because check names are not unique, we must
// also keep track of the counts on each side and make sure those also match.
services := tg.ConsulServices()
expChecks := make(map[string]int)
regChecks := make(map[string]int)
for _, service := range services {
Expand Down