Merge pull request #846 from jshufro/jms/constraintscleanup

Clean up some version specific logic for versions well behind upgrade barriers

Author: 0xfornax

rocketpool-cli/service/service.go

@@ -3,13 +3,11 @@ package service
 import (
 	"fmt"
 	"os"
-	"os/exec"
 	"path/filepath"
 	"regexp"
 	"strings"
 	"time"
 
-	"github.com/hashicorp/go-version"
 	"github.com/mitchellh/go-homedir"
 	"github.com/rivo/tview"
 	"github.com/urfave/cli"
@@ -556,45 +554,6 @@ func startService(c *cli.Context, ignoreConfigSuggestion bool) error {
 		fmt.Printf("%sIgnoring anti-slashing safety delay.%s\n", colorYellow, colorReset)
 	}
 
-	// Force a delay if using Teku and upgrading from v1.3.0 or below because of the slashing protection DB migration in v1.3.1+
-	isLocalTeku := (cfg.ConsensusClientMode.Value.(cfgtypes.Mode) == cfgtypes.Mode_Local && cfg.ConsensusClient.Value.(cfgtypes.ConsensusClient) == cfgtypes.ConsensusClient_Teku)
-	isExternalTeku := (cfg.ConsensusClientMode.Value.(cfgtypes.Mode) == cfgtypes.Mode_External && cfg.ExternalConsensusClient.Value.(cfgtypes.ConsensusClient) == cfgtypes.ConsensusClient_Teku)
-	if isUpdate && !isNew && !cfg.IsNativeMode && (isLocalTeku || isExternalTeku) && !c.Bool("ignore-slash-timer") {
-		previousVersion := "0.0.0"
-		backupCfg, err := rp.LoadBackupConfig()
-		if err != nil {
-			fmt.Printf("WARNING: Couldn't determine previous Smart Node version from backup settings: %s\n", err.Error())
-		} else if backupCfg != nil {
-			previousVersion = backupCfg.Version
-		}
-
-		oldVersion, err := version.NewVersion(strings.TrimPrefix(previousVersion, "v"))
-		if err != nil {
-			fmt.Printf("WARNING: Backup configuration states the previous Smart Node installation used version %s, which is not a valid version\n", previousVersion)
-			oldVersion, _ = version.NewVersion("0.0.0")
-		}
-
-		vulnerableConstraint, _ := version.NewConstraint("<= 1.3.0")
-		if vulnerableConstraint.Check(oldVersion) {
-			err = handleTekuSlashProtectionMigrationDelay(rp, cfg)
-			if err != nil {
-				return err
-			}
-		}
-	}
-
-	// Force stop eth2 if using Nimbus prior to v1.8.0 so it ensures the container is shut down and thus lets go of the validator keys and slashing database
-	isLocalNimbus := (cfg.ConsensusClientMode.Value.(cfgtypes.Mode) == cfgtypes.Mode_Local && cfg.ConsensusClient.Value.(cfgtypes.ConsensusClient) == cfgtypes.ConsensusClient_Nimbus)
-	if isUpdate && !isNew && !cfg.IsNativeMode && isLocalNimbus {
-		proceed, err := handleNimbusSplitConversion(rp, cfg)
-		if err != nil {
-			return fmt.Errorf("error handling Nimbus split-mode upgrade: %w", err)
-		}
-		if !proceed {
-			return nil
-		}
-	}
-
 	// Write a note on doppelganger protection
 	doppelgangerEnabled, err := cfg.IsDoppelgangerEnabled()
 	if err != nil {
@@ -614,182 +573,6 @@ func startService(c *cli.Context, ignoreConfigSuggestion bool) error {
 
 }
 
-// Versions prior to v1.9.0 had Nimbus in single mode instead of split mode, so handle the conversion to ensure the user doesn't get slashed
-func handleNimbusSplitConversion(rp *rocketpool.Client, cfg *config.RocketPoolConfig) (bool, error) {
-
-	previousVersion := "0.0.0"
-	backupCfg, err := rp.LoadBackupConfig()
-	if err != nil {
-		fmt.Printf("%sWARNING: Couldn't determine previous Smart Node version from backup settings: %s%s\n", colorYellow, err.Error(), colorReset)
-		fmt.Printf("%sYou are configured to use Nimbus in local mode. Starting with v1.9.0, Nimbus is now configured to use a split-process configuration, which means the Beacon Node (the `eth2` container) no longer loads your validator keys - now the `validator` container does.\n\nDue to this, we must restart Nimbus as part of the upgrade.\n\nIf you were previously running Smart Node v1.7.5 or earlier, you **MUST** shut down the Docker containers with `rocketpool service stop` and wait **at least 15 minutes** to ensure that you've missed at least two attestations before proceeding to prevent being slashed. Please use an explorer such as https://beaconcha.in to confirm at least one of the missed attestations has been finalized before proceeding.%s\n\n", colorYellow, colorReset)
-		fmt.Println()
-		if !prompt.Confirm(fmt.Sprintf("Press y when you understand the above warning, have waited, and are ready to start Rocket Pool:%s", colorReset)) {
-			fmt.Println("Cancelled.")
-			return false, nil
-		}
-		return true, nil
-	} else if backupCfg != nil {
-		previousVersion = backupCfg.Version
-	} else {
-		fmt.Printf("%sWARNING: Couldn't determine previous Smart Node version from backup settings because the backup configuration didn't exist.%s\n", colorYellow, colorReset)
-		fmt.Printf("%sYou are configured to use Nimbus in local mode. Starting with v1.9.0, Nimbus is now configured to use a split-process configuration, which means the Beacon Node (the `eth2` container) no longer loads your validator keys - now the `validator` container does.\n\nDue to this, we must restart Nimbus as part of the upgrade.\n\nIf you were previously running Smart Node v1.7.5 or earlier, you **MUST** shut down the Docker containers with `rocketpool service stop` and wait **at least 15 minutes** to ensure that you've missed at least two attestations before proceeding to prevent being slashed. Please use an explorer such as https://beaconcha.in to confirm at least one of the missed attestations has been finalized before proceeding.%s\n\n", colorYellow, colorReset)
-		fmt.Println()
-		if !prompt.Confirm(fmt.Sprintf("Press y when you understand the above warning, have waited, and are ready to start Rocket Pool:%s", colorReset)) {
-			fmt.Println("Cancelled.")
-			return false, nil
-		}
-		return true, nil
-	}
-
-	oldVersion, err := version.NewVersion(strings.TrimPrefix(previousVersion, "v"))
-	if err != nil {
-		fmt.Printf("%sWARNING: Backup configuration states the previous Smart Node installation used version %s, which is not a valid version%s\n", colorYellow, previousVersion, colorReset)
-		fmt.Printf("%sYou are configured to use Nimbus in local mode. Starting with v1.9.0, Nimbus is now configured to use a split-process configuration, which means the Beacon Node (the `eth2` container) no longer loads your validator keys - now the `validator` container does.\n\nDue to this, we must restart Nimbus as part of the upgrade.\n\nIf you were previously running Smart Node v1.7.5 or earlier, you **MUST** shut down the Docker containers with `rocketpool service stop` and wait **at least 15 minutes** to ensure that you've missed at least two attestations before proceeding to prevent being slashed. Please use an explorer such as https://beaconcha.in to confirm at least one of the missed attestations has been finalized before proceeding.%s\n\n", colorYellow, colorReset)
-		fmt.Println()
-		if !prompt.Confirm(fmt.Sprintf("Press y when you understand the above warning, have waited, and are ready to start Rocket Pool:%s", colorReset)) {
-			fmt.Println("Cancelled.")
-			return false, nil
-		}
-		return true, nil
-	}
-
-	vulnerableConstraint, _ := version.NewConstraint("< 1.8.0")
-	if vulnerableConstraint.Check(oldVersion) {
-		fmt.Println()
-		fmt.Printf("%sNOTE: You are configured to use Nimbus in local mode. Starting with v1.9.0, Nimbus is now configured to use a split-process configuration, which means the Beacon Node (the `eth2` container) no longer loads your validator keys - now the `validator` container does.\n\nDue to this, we must restart Nimbus as part of the upgrade. Your client's slashing database will be moved from the `eth2` container to the `validator` container automatically to ensure your node doesn't attest to the same duty twice and get slashed.\n\nIf you have *any concern at all* about this process, you may want to voluntarily shut down the Docker containers with `rocketpool service stop` and wait 15 minutes to ensure that you've missed at least two attestations before proceeding. If you do this, please use an explorer such as https://beaconcha.in to confirm at least one of the missed attestations has been finalized before proceeding.%s\n\n", colorYellow, colorReset)
-		fmt.Println()
-		if !prompt.Confirm("Do you want to continue starting the service?") {
-			fmt.Println("Cancelled.")
-			return false, nil
-		}
-
-		// Ensure the eth2 and validator containers have stopped
-		prefix, err := rp.GetContainerPrefix()
-		if err != nil {
-			return false, fmt.Errorf("error getting container prefix: %w", err)
-		}
-
-		successfulStop := true
-		eth2ContainerName := prefix + BeaconContainerSuffix
-		fmt.Printf("Stopping %s...\n", eth2ContainerName)
-		out, err := rp.StopContainer(eth2ContainerName)
-		if err != nil {
-			exitErr, isExitErr := err.(*exec.ExitError)
-			if isExitErr && exitErr.ProcessState.ExitCode() == 1 && strings.Contains(string(exitErr.Stderr), "No such container:") {
-				// Handle errors where the container didn't exist
-				fmt.Printf("%sNOTE: couldn't shut down %s because it didn't exist.%s\n", colorYellow, eth2ContainerName, colorReset)
-				successfulStop = false
-			} else {
-				return false, fmt.Errorf("error stopping %s: %w", eth2ContainerName, err)
-			}
-		} else if out != eth2ContainerName {
-			return false, fmt.Errorf("unexpected output when trying to stop %s: [%s]", eth2ContainerName, out)
-		}
-
-		validatorContainerName := prefix + ValidatorContainerSuffix
-		fmt.Printf("Stopping %s...\n", validatorContainerName)
-		out, err = rp.StopContainer(validatorContainerName)
-		if err != nil {
-			exitErr, isExitErr := err.(*exec.ExitError)
-			if isExitErr && exitErr.ProcessState.ExitCode() == 1 && strings.Contains(string(exitErr.Stderr), "No such container:") {
-				// Handle errors where the container didn't exist
-				fmt.Printf("%sNOTE: couldn't shut down %s because it didn't exist.%s\n", colorYellow, validatorContainerName, colorReset)
-				successfulStop = false
-			} else {
-				return false, fmt.Errorf("error stopping %s: %w", validatorContainerName, err)
-			}
-		} else if out != validatorContainerName {
-			return false, fmt.Errorf("unexpected output when trying to stop %s: [%s]", validatorContainerName, out)
-		}
-
-		if !successfulStop {
-			fmt.Println()
-			fmt.Printf("%sWARNING: Some of the Nimbus containers couldn't be shut down safely.\nThe Smart Node can't guarantee the safe transfer of the slashing database. If you have active validators, you **must ensure** you have waited 15 minutes since your last attestation and **missed at least two attestations** before continuing.\nIf you don't, you %sMAY BE SLASHED!%s\n\n", colorYellow, colorRed, colorReset)
-			fmt.Println()
-			if !prompt.Confirm(fmt.Sprintf("Press y when you understand the above warning, have waited, and are ready to start Rocket Pool:%s", colorReset)) {
-				fmt.Println("Cancelled.")
-				return false, nil
-			}
-		}
-	}
-
-	return true, nil
-
-}
-
-// Versions prior to v1.3.1 didn't preserve Teku's slashing DB, so force a delay when upgrading to ensure the user doesn't get slashed by accident
-func handleTekuSlashProtectionMigrationDelay(rp *rocketpool.Client, cfg *config.RocketPoolConfig) error {
-
-	fmt.Printf("%s=== NOTICE ===\n", colorYellow)
-	fmt.Printf("You are currently using Teku as your Consensus client.\nv1.3.1+ fixes an issue that would cause Teku's slashing protection database to be lost after an upgrade.\nIt will now be rebuilt.\n\nFor the absolute safety of your funds, your node will wait for 15 minutes before starting.\nYou will miss a few attestations during this process; this is expected.\n\nThis delay only needs to happen the first time you start the Smart Node after upgrading to v1.3.1 or higher.%s\n\nIf you are installing the Smart Node for the first time or don't have any validators yet, you can skip this with `rocketpool service start --ignore-slash-timer`. Otherwise, we strongly recommend you wait for the full delay.\n\n", colorReset)
-
-	// Get the container prefix
-	prefix, err := rp.GetContainerPrefix()
-	if err != nil {
-		return fmt.Errorf("Error getting validator container prefix: %w", err)
-	}
-
-	// Get the current validator client
-	currentValidatorImageString, err := rp.GetDockerImage(prefix + ValidatorContainerSuffix)
-	if err != nil {
-		return fmt.Errorf("Error getting current validator image: %w", err)
-	}
-
-	currentValidatorName, err := getDockerImageName(currentValidatorImageString)
-	if err != nil {
-		return fmt.Errorf("Error getting current validator image name: %w", err)
-	}
-
-	// Get the time that the container responsible for validator duties exited
-	validatorDutyContainerName, err := getContainerNameForValidatorDuties(currentValidatorName, rp)
-	if err != nil {
-		return fmt.Errorf("Error getting validator container name: %w", err)
-	}
-	validatorFinishTime, err := rp.GetDockerContainerShutdownTime(validatorDutyContainerName)
-	if err != nil {
-		return fmt.Errorf("Error getting validator shutdown time: %w", err)
-	}
-
-	// If it hasn't exited yet, shut it down
-	zeroTime := time.Time{}
-	status, err := rp.GetDockerStatus(validatorDutyContainerName)
-	if err != nil {
-		return fmt.Errorf("Error getting container [%s] status: %w", validatorDutyContainerName, err)
-	}
-	if validatorFinishTime == zeroTime || status == "running" {
-		fmt.Printf("%sValidator is currently running, stopping it...%s\n", colorYellow, colorReset)
-		response, err := rp.StopContainer(validatorDutyContainerName)
-		validatorFinishTime = time.Now()
-		if err != nil {
-			return fmt.Errorf("Error stopping container [%s]: %w", validatorDutyContainerName, err)
-		}
-		if response != validatorDutyContainerName {
-			return fmt.Errorf("Unexpected response when stopping container [%s]: %s", validatorDutyContainerName, response)
-		}
-	}
-
-	// Print the warning and start the time lockout
-	safeStartTime := validatorFinishTime.Add(15 * time.Minute)
-	remainingTime := time.Until(safeStartTime)
-	if remainingTime <= 0 {
-		fmt.Printf("The validator has been offline for %s, which is long enough to prevent slashing.\n", time.Since(validatorFinishTime))
-		fmt.Println("The new client can be safely started.")
-	} else {
-		// Wait for 15 minutes
-		for remainingTime > 0 {
-			fmt.Printf("Remaining time: %s", remainingTime)
-			time.Sleep(1 * time.Second)
-			remainingTime = time.Until(safeStartTime)
-			fmt.Printf("%s\r", clearLine)
-		}
-
-		fmt.Println(colorReset)
-		fmt.Println("You may now safely start the validator without fear of being slashed.")
-	}
-
-	return nil
-}
-
 func checkForValidatorChange(rp *rocketpool.Client, cfg *config.RocketPoolConfig) error {
 
 	// Get the container prefix
@@ -824,9 +607,6 @@ func checkForValidatorChange(rp *rocketpool.Client, cfg *config.RocketPoolConfig
 		fmt.Printf("Validator client [%s] was previously used - no slashing prevention delay necessary.\n", currentValidatorName)
 	} else if currentValidatorName == "" {
 		fmt.Println("This is the first time starting Rocket Pool - no slashing prevention delay necessary.")
-	} else if (currentValidatorName == "nimbus-eth2" && pendingValidatorName == "nimbus-validator-client") || (pendingValidatorName == "nimbus-eth2" && currentValidatorName == "nimbus-validator-client") {
-		// Handle the transition from Nimbus v22.11.x to Nimbus v22.12.x where they split the VC into its own container
-		fmt.Printf("Validator client [%s] was previously used, you are changing to [%s] but the Smart Node will migrate your slashing database automatically to this new client. No slashing prevention delay is necessary.\n", currentValidatorName, pendingValidatorName)
 	} else {
 
 		consensusClient, _ := cfg.GetSelectedConsensusClient()