open-nomad/command/agent/consul/script.go
Michael Schurter 8fa5e90095 consul: add ScriptExecutor context wrapper
Since d335a82859ca2177bc6deda0c2c85b559daf2db3 ScriptExecutors now take
a timeout duration instead of a context. This broke the script check
removal code which used context cancelation propagation to remove
script checks while they were executing.

This commit adds a wrapper around ScriptExecutors that obeys context
cancelation again. The only downside is that it leaks a goroutine until
the underlying Exec call completes or timeouts.

Since check removal is relatively rare, check timeouts usually low, and
scripts usually fast, the risk of leaking a goroutine seems very small.
2018-12-03 20:26:31 -08:00

216 lines
5.5 KiB
Go

package consul
import (
"context"
"time"
metrics "github.com/armon/go-metrics"
log "github.com/hashicorp/go-hclog"
"github.com/hashicorp/consul/api"
"github.com/hashicorp/nomad/client/allocrunner/taskrunner/interfaces"
"github.com/hashicorp/nomad/nomad/structs"
)
// heartbeater is the subset of consul agent functionality needed by script
// checks to heartbeat
type heartbeater interface {
UpdateTTL(id, output, status string) error
}
// contextExec allows canceling a ScriptExecutor with a context.
type contextExec struct {
// pctx is the parent context. A subcontext will be created with Exec's
// timeout.
pctx context.Context
// exec to be wrapped in a context
exec interfaces.ScriptExecutor
}
func newContextExec(ctx context.Context, exec interfaces.ScriptExecutor) *contextExec {
return &contextExec{
pctx: ctx,
exec: exec,
}
}
type execResult struct {
buf []byte
code int
err error
}
// Exec a command until the timeout expires, the context is canceled, or the
// underlying Exec returns.
func (c *contextExec) Exec(timeout time.Duration, cmd string, args []string) ([]byte, int, error) {
resCh := make(chan execResult, 1)
// Don't trust the underlying implementation to obey timeout
ctx, cancel := context.WithTimeout(c.pctx, timeout)
defer cancel()
go func() {
output, code, err := c.exec.Exec(timeout, cmd, args)
select {
case resCh <- execResult{output, code, err}:
case <-ctx.Done():
}
}()
select {
case res := <-resCh:
return res.buf, res.code, res.err
case <-ctx.Done():
return nil, 0, ctx.Err()
}
}
// scriptHandle is returned by scriptCheck.run by cancelling a scriptCheck and
// waiting for it to shutdown.
type scriptHandle struct {
// cancel the script
cancel func()
exitCh chan struct{}
}
// wait returns a chan that's closed when the script exits
func (s *scriptHandle) wait() <-chan struct{} {
return s.exitCh
}
// scriptCheck runs script checks via a ScriptExecutor and updates the
// appropriate check's TTL when the script succeeds.
type scriptCheck struct {
allocID string
taskName string
id string
check *structs.ServiceCheck
exec interfaces.ScriptExecutor
agent heartbeater
// lastCheckOk is true if the last check was ok; otherwise false
lastCheckOk bool
logger log.Logger
shutdownCh <-chan struct{}
}
// newScriptCheck creates a new scriptCheck. run() should be called once the
// initial check is registered with Consul.
func newScriptCheck(allocID, taskName, checkID string, check *structs.ServiceCheck,
exec interfaces.ScriptExecutor, agent heartbeater, logger log.Logger,
shutdownCh <-chan struct{}) *scriptCheck {
logger = logger.ResetNamed("consul.checks").With("task", taskName, "alloc_id", allocID, "check", check.Name)
return &scriptCheck{
allocID: allocID,
taskName: taskName,
id: checkID,
check: check,
exec: exec,
agent: agent,
lastCheckOk: true, // start logging on first failure
logger: logger,
shutdownCh: shutdownCh,
}
}
// run this script check and return its cancel func. If the shutdownCh is
// closed the check will be run once more before exiting.
func (s *scriptCheck) run() *scriptHandle {
ctx, cancel := context.WithCancel(context.Background())
exitCh := make(chan struct{})
// Wrap the original ScriptExecutor in one that obeys context
// cancelation.
ctxExec := newContextExec(ctx, s.exec)
go func() {
defer close(exitCh)
timer := time.NewTimer(0)
defer timer.Stop()
for {
// Block until check is removed, Nomad is shutting
// down, or the check interval is up
select {
case <-ctx.Done():
// check has been removed
return
case <-s.shutdownCh:
// unblock but don't exit until after we heartbeat once more
case <-timer.C:
timer.Reset(s.check.Interval)
}
metrics.IncrCounter([]string{"client", "consul", "script_runs"}, 1)
// Execute check script with timeout
output, code, err := ctxExec.Exec(s.check.Timeout, s.check.Command, s.check.Args)
switch err {
case context.Canceled:
// check removed during execution; exit
return
case context.DeadlineExceeded:
metrics.IncrCounter([]string{"client", "consul", "script_timeouts"}, 1)
// If no error was returned, set one to make sure the task goes critical
if err == nil {
err = context.DeadlineExceeded
}
// Log deadline exceeded every time as it's a
// distinct issue from checks returning
// failures
s.logger.Warn("check timed out", "timeout", s.check.Timeout)
}
state := api.HealthCritical
switch code {
case 0:
state = api.HealthPassing
case 1:
state = api.HealthWarning
}
var outputMsg string
if err != nil {
state = api.HealthCritical
outputMsg = err.Error()
} else {
outputMsg = string(output)
}
// Actually heartbeat the check
err = s.agent.UpdateTTL(s.id, outputMsg, state)
select {
case <-ctx.Done():
// check has been removed; don't report errors
return
default:
}
if err != nil {
if s.lastCheckOk {
s.lastCheckOk = false
s.logger.Warn("updating check failed", "error", err)
} else {
s.logger.Debug("updating check still failing", "error", err)
}
} else if !s.lastCheckOk {
// Succeeded for the first time or after failing; log
s.lastCheckOk = true
s.logger.Info("updating check succeeded")
}
select {
case <-s.shutdownCh:
// We've been told to exit and just heartbeated so exit
return
default:
}
}
}()
return &scriptHandle{cancel: cancel, exitCh: exitCh}
}