Merge pull request #1392 from hashicorp/maint_mode_timeout

Change the way we set the timeout for maintenance mode

CHANGELOG.md

@@ -2,10 +2,11 @@
 
 BUG FIXES:
 * Minor fixes of issues that came up during testing against vSphere 7.0
+* Change the way we set the timeout for maintenance mode ([#1392](https://github.com/hashicorp/terraform-provider-vsphere/pull/1392)) 
 
 IMPROVEMENTS:
-* `provider`: vSphere 7 compatibility validation ([1381](https://github.com/hashicorp/terraform-provider-vsphere/pull/1381))
-* `resource/vm`: Allow hardware version up to 19 ([1391](https://github.com/hashicorp/terraform-provider-vsphere/pull/1391))
+* `provider`: vSphere 7 compatibility validation ([#1381](https://github.com/hashicorp/terraform-provider-vsphere/pull/1381))
+* `resource/vm`: Allow hardware version up to 19 ([#1391](https://github.com/hashicorp/terraform-provider-vsphere/pull/1391))
 
 ## 1.25.0 (March 17, 2021)
 

vsphere/internal/helper/clustercomputeresource/cluster_compute_resource_helper.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"log"
 	"strings"
+	"time"
 
 	"github.com/hashicorp/terraform-provider-vsphere/vsphere/internal/helper/computeresource"
 	"github.com/hashicorp/terraform-provider-vsphere/vsphere/internal/helper/folder"
@@ -231,7 +232,8 @@ func MoveHostsInto(client *govmomi.Client, cluster *object.ClusterComputeResourc
 				}
 			}
 
-			err = hostsystem.EnterMaintenanceMode(hs, int(provider.DefaultAPITimeout.Seconds())+(int(provider.DefaultAPITimeout.Seconds())*len(hsProps.Vm)), evacuate)
+			totalVmTimeout := provider.DefaultAPITimeout * time.Duration(len(hsProps.Vm)+1)
+			err = hostsystem.EnterMaintenanceMode(hs, totalVmTimeout, evacuate)
 			if err != nil {
 				return fmt.Errorf("while putting host %q in maintenance mode: %s", hs.Reference().Value, err)
 			}
@@ -279,7 +281,8 @@ func MoveHostsOutOf(cluster *object.ClusterComputeResource, hosts []*object.Host
 
 func moveHostOutOf(cluster *object.ClusterComputeResource, host *object.HostSystem, timeout int) error {
 	// Place the host into maintenance mode. This blocks until the host is ready.
-	if err := hostsystem.EnterMaintenanceMode(host, timeout, true); err != nil {
+	timeoutDuration := time.Duration(timeout) * time.Second
+	if err := hostsystem.EnterMaintenanceMode(host, timeoutDuration, true); err != nil {
 		return fmt.Errorf("error putting host %q into maintenance mode: %s", host.Name(), err)
 	}
 
@@ -294,7 +297,7 @@ func moveHostOutOf(cluster *object.ClusterComputeResource, host *object.HostSyst
 	}
 
 	// Move the host out of maintenance mode now that it's out of the cluster.
-	if err := hostsystem.ExitMaintenanceMode(host, timeout); err != nil {
+	if err := hostsystem.ExitMaintenanceMode(host, timeoutDuration); err != nil {
 		return fmt.Errorf("error taking host %q out of maintenance mode: %s", host.Name(), err)
 	}
 

vsphere/internal/helper/hostsystem/host_system_helper.go

@@ -114,7 +114,7 @@ func HostInMaintenance(host *object.HostSystem) (bool, error) {
 // to true, all powered off VMs will be removed from the host, or the task will
 // block until this is the case, depending on whether or not DRS is on or off
 // for the host's cluster. This parameter is ignored on direct ESXi.
-func EnterMaintenanceMode(host *object.HostSystem, timeout int, evacuate bool) error {
+func EnterMaintenanceMode(host *object.HostSystem, timeout time.Duration, evacuate bool) error {
 	if err := viapi.VimValidateVirtualCenter(host.Client()); err != nil {
 		evacuate = false
 	}
@@ -127,9 +127,9 @@ func EnterMaintenanceMode(host *object.HostSystem, timeout int, evacuate bool) e
 
 	log.Printf("[DEBUG] Host %q is entering maintenance mode (evacuate: %t)", host.Name(), evacuate)
 
-	ctx, cancel := context.WithTimeout(context.Background(), time.Second*time.Duration(timeout))
+	ctx, cancel := context.WithTimeout(context.Background(), timeout)
 	defer cancel()
-	task, err := host.EnterMaintenanceMode(ctx, int32(timeout), evacuate, nil)
+	task, err := host.EnterMaintenanceMode(ctx, int32(timeout.Seconds()), evacuate, nil)
 	if err != nil {
 		return err
 	}
@@ -152,7 +152,7 @@ func EnterMaintenanceMode(host *object.HostSystem, timeout int, evacuate bool) e
 }
 
 // ExitMaintenanceMode takes a host out of maintenance mode.
-func ExitMaintenanceMode(host *object.HostSystem, timeout int) error {
+func ExitMaintenanceMode(host *object.HostSystem, timeout time.Duration) error {
 	maintMode, err := HostInMaintenance(host)
 	if !maintMode {
 		log.Printf("[DEBUG] Host %q is already not in maintenance mode", host.Name())
@@ -161,13 +161,9 @@ func ExitMaintenanceMode(host *object.HostSystem, timeout int) error {
 
 	log.Printf("[DEBUG] Host %q is exiting maintenance mode", host.Name())
 
-	// Add 5 minutes to timeout for the context timeout to allow for any issues
-	// with the request after.
-	// TODO: Fix this so that it ultimately uses the provider context.
-	ctxTimeout := timeout + 300
-	ctx, cancel := context.WithTimeout(context.Background(), time.Second*time.Duration(ctxTimeout))
+	ctx, cancel := context.WithTimeout(context.Background(), timeout)
 	defer cancel()
-	task, err := host.ExitMaintenanceMode(ctx, int32(timeout))
+	task, err := host.ExitMaintenanceMode(ctx, int32(timeout.Seconds()))
 	if err != nil {
 		return err
 	}

vsphere/resource_vsphere_compute_cluster.go

@@ -153,10 +153,11 @@ func resourceVSphereComputeCluster() *schema.Resource {
 				StateFunc:   folder.NormalizePath,
 			},
 			"host_cluster_exit_timeout": {
-				Type:        schema.TypeInt,
-				Optional:    true,
-				Default:     3600,
-				Description: "The timeout for each host maintenance mode operation when removing hosts from a cluster.",
+				Type:         schema.TypeInt,
+				Optional:     true,
+				Default:      3600,
+				Description:  "The timeout for each host maintenance mode operation when removing hosts from a cluster.",
+				ValidateFunc: validation.IntBetween(0, 604800),
 			},
 			"force_evacuate_on_destroy": {
 				Type:        schema.TypeBool,
@@ -806,7 +807,7 @@ func resourceVSphereComputeClusterProcessHostUpdate(
 				return fmt.Errorf("while fetching properties for host %q: %s", hs.Reference().Value, err)
 			}
 			if hsProps.Runtime.InMaintenanceMode {
-				err := hostsystem.ExitMaintenanceMode(hs, int(provider.DefaultAPITimeout))
+				err := hostsystem.ExitMaintenanceMode(hs, provider.DefaultAPITimeout)
 				if err != nil {
 					return fmt.Errorf("while getting host %q out of maintenance mode: %s", hs.Reference().Value, err)
 				}

vsphere/resource_vsphere_host.go

@@ -4,7 +4,8 @@ import (
 	"context"
 	"fmt"
 	"log"
-	"time"
+
+	"github.com/hashicorp/terraform-provider-vsphere/vsphere/internal/helper/provider"
 
 	"github.com/hashicorp/terraform-plugin-sdk/helper/schema"
 	"github.com/hashicorp/terraform-plugin-sdk/helper/validation"
@@ -294,9 +295,9 @@ func resourceVsphereHostCreate(d *schema.ResourceData, meta interface{}) error {
 
 	maintenanceMode := d.Get("maintenance").(bool)
 	if maintenanceMode {
-		err = hostsystem.EnterMaintenanceMode(host, int(defaultAPITimeout/time.Minute), true)
+		err = hostsystem.EnterMaintenanceMode(host, provider.DefaultAPITimeout, true)
 	} else {
-		err = hostsystem.ExitMaintenanceMode(host, int(defaultAPITimeout/time.Minute))
+		err = hostsystem.ExitMaintenanceMode(host, provider.DefaultAPITimeout)
 	}
 	if err != nil {
 		return fmt.Errorf("error while toggling maintenance mode for host %s. Error: %s", hostID, err)
@@ -478,9 +479,9 @@ func resourceVSphereHostUpdateMaintenanceMode(d *schema.ResourceData, meta, old,
 
 	maintenanceMode := newVal.(bool)
 	if maintenanceMode {
-		err = hostsystem.EnterMaintenanceMode(host, int(defaultAPITimeout/time.Minute), true)
+		err = hostsystem.EnterMaintenanceMode(host, provider.DefaultAPITimeout, true)
 	} else {
-		err = hostsystem.ExitMaintenanceMode(host, int(defaultAPITimeout/time.Minute))
+		err = hostsystem.ExitMaintenanceMode(host, provider.DefaultAPITimeout)
 	}
 	if err != nil {
 		return fmt.Errorf("error while toggling maintenance mode for host %s. Error: %s", host.Name(), err)
@@ -517,7 +518,7 @@ func resourceVSphereHostUpdateCluster(d *schema.ResourceData, meta, old, newVal
 		return fmt.Errorf("error while retrieving HostSystem object for host ID %s. Error: %s", hostID, err)
 	}
 
-	err = hostsystem.EnterMaintenanceMode(hs, int(defaultAPITimeout/time.Minute), true)
+	err = hostsystem.EnterMaintenanceMode(hs, provider.DefaultAPITimeout, true)
 	if err != nil {
 		return fmt.Errorf("error while putting host to maintenance mode: %s", err.Error())
 	}
@@ -532,7 +533,7 @@ func resourceVSphereHostUpdateCluster(d *schema.ResourceData, meta, old, newVal
 		return fmt.Errorf("error while moving host to new cluster (%s): %s", newClusterID, err)
 	}
 
-	err = hostsystem.ExitMaintenanceMode(hs, int(defaultAPITimeout/time.Minute))
+	err = hostsystem.ExitMaintenanceMode(hs, provider.DefaultAPITimeout)
 	if err != nil {
 		return fmt.Errorf("error while taking host out of maintenance mode: %s", err.Error())
 	}
@@ -568,7 +569,7 @@ func resourceVSphereHostReconnect(d *schema.ResourceData, meta interface{}) erro
 
 	maintenanceConfig := d.Get("maintenance").(bool)
 	if maintenanceState && !maintenanceConfig {
-		err := hostsystem.ExitMaintenanceMode(host, int(defaultAPITimeout/time.Minute))
+		err := hostsystem.ExitMaintenanceMode(host, provider.DefaultAPITimeout)
 		if err != nil {
 			return fmt.Errorf("error while taking host %s out of maintenance mode. Error: %s", host.Name(), err)
 		}