and also some cluster/scripts/ while we're at it

s.a. #1032

[ci]

Signed-off-by: Martin Florian <[email protected]>

Author: martinflorian-da

cluster/scripts/find-recent-backup.sh

@@ -0,0 +1,152 @@
+#!/usr/bin/env bash
+
+# Copyright (c) 2024 Digital Asset (Switzerland) GmbH and/or its affiliates. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+set -euo pipefail
+
+# shellcheck disable=SC1091
+source "${TOOLS_LIB}/libcli.source"
+# shellcheck disable=SC1091
+source "${SPLICE_ROOT}/cluster/scripts/utils.source"
+
+function usage() {
+  _info "Usage: $0 <namespace> <migration_id> <internal (true|false)>"
+}
+
+function is_full_backup_kube() {
+  local component_backup_names=$1
+  local expected_components=$2
+
+  # Check if all expected components can be found in the component_backup_names
+  for component in $expected_components; do
+    count=$(echo "$component_backup_names" | grep -c "$component")
+    if [ "$count" -ne 1 ]; then
+      return 1
+    fi
+  done
+
+  return 0
+}
+
+function get_component_backup_names_kube() {
+  local migration_id=$1
+  local run_id=$2
+  component_backup_names=$(kubectl get volumesnapshot -n "$namespace" --sort-by=.metadata.creationTimestamp -o json | jq "[.items[] | select(.metadata.annotations[\"migrationId\"] == \"$migration_id\") | select(.metadata.name | endswith(\"$run_id\"))]" | jq -r '.[].metadata.name // empty' )
+  echo "$component_backup_names"
+}
+
+function latest_full_backup_run_id_kube() {
+  local namespace=$1
+  local migration_id=$2
+  local is_sv=$3
+  local expected_components=$4
+  if [ "$is_sv" == "true" ]; then
+      expected_components="$expected_components cometbft"
+  fi
+
+  local all_run_ids
+  # get all run id postgres
+  all_run_ids=$(kubectl get volumesnapshot -n "$namespace" --sort-by=.metadata.creationTimestamp -o json | jq "[.items[] | select(.metadata.annotations[\"migrationId\"] == \"$migration_id\")]" | jq -r '.[].metadata.name // empty' | grep -o '[^-]*$' | sort -rn | uniq )
+
+  while read -r run_id; do
+    component_backup_names=$(get_component_backup_names_kube "$migration_id" "$run_id")
+    if is_full_backup_kube "$component_backup_names" "$expected_components"; then
+      echo "$run_id"
+      return 0
+    fi
+  done <<< "$all_run_ids"
+  return 1
+}
+
+function latest_full_backup_run_id_gcloud() {
+  local namespace=$1
+  local migration_id=$2
+  local is_sv=$3
+  local expected_components=$4
+  local internal=$5
+  local num_components
+  num_components=$(echo "$expected_components" | wc -w)
+  local stack
+
+  declare -A run_ids_dict
+
+  for component in $expected_components; do
+    stack=$(get_stack_for_namespace_component "$namespace" "$component" "$internal")
+    instance="$(create_component_instance "$component" "$migration_id" "$namespace" "$internal")"
+    local full_component_instance="$namespace-$instance-pg"
+
+    local cloudsql_id
+    cloudsql_id=$(get_cloudsql_id "$full_component_instance" "$stack")
+    # We always create backups with a description field while this field could be missing for automated backups done by Google Cloud.
+    # So we filter those out while looking for the most recent backup.
+    mapfile -t run_ids < <(gcloud sql backups list --instance "$cloudsql_id" --filter=type="ON_DEMAND" --format=json | jq -r '.[] | select(has("description")) | .description')
+    run_ids_dict[$component]="${run_ids[*]}"
+  done
+
+  if [ "$is_sv" == "true" ]; then
+    mapfile -t cometbft_run_ids < <(kubectl get volumesnapshot -n "$namespace" --sort-by=.metadata.creationTimestamp -o json | jq "[.items[] | select(.metadata.annotations[\"migrationId\"] == \"$migration_id\") | select(.metadata.name | contains(\"cometbft\"))]" | jq -r '.[].metadata.name // empty' | grep -o '[^-]*$' | sort -rn | uniq )
+    run_ids_dict["cometbft"]="${cometbft_run_ids[*]}"
+    ((num_components++))
+  fi
+
+  declare -A count_dict
+  for key in "${!run_ids_dict[@]}"; do
+    for value in ${run_ids_dict[$key]}; do
+      if [ -z "${count_dict[$value]+_}" ]; then
+        count_dict[$value]=0
+      fi
+      ((count_dict[$value]++))
+    done
+  done
+
+  largest_run_id=$(for key in "${!count_dict[@]}"; do
+    if [ "${count_dict[$key]}" -eq "$num_components" ]; then
+      echo "$key"
+    fi
+  done | sort -n | tail -n 1)
+
+  echo "$largest_run_id"
+}
+
+function main() {
+  if [ "$#" -lt 3 ]; then
+      usage
+      exit 1
+  fi
+
+  local namespace=$1
+  local migration_id=$2
+  local internal=$3
+
+  case "$namespace" in
+      sv-1|sv-2|sv-3|sv-4)
+          is_sv=true
+          full_instance="$namespace-cn-apps-pg"
+          expected_components="cn-apps sequencer participant mediator"
+          stack=$(get_stack_for_namespace_component "$namespace" "cn-apps" "$internal")
+          ;;
+      *)
+          is_sv=false
+          full_instance="$namespace-validator-pg"
+          expected_components="validator participant"
+          stack=$(get_stack_for_namespace_component "$namespace" "participant" "$internal")
+          ;;
+  esac
+
+  type=$(get_postgres_type "$full_instance" "$stack")
+  # We only check the postgres type of one component and assume other components have the same type.
+  if [ "$type" == "canton:network:postgres" ]; then
+    backup_run_id=$(latest_full_backup_run_id_kube "$namespace" "$migration_id" "$is_sv" "$expected_components")
+    echo "$backup_run_id"
+  elif [ "$type" == "canton:cloud:postgres" ]; then
+    backup_run_id=$(latest_full_backup_run_id_gcloud "$namespace" "$migration_id" "$is_sv" "$expected_components" "$internal")
+    echo "$backup_run_id"
+  elif [ -z "$type" ]; then
+    _error "No postgres instance $full_instance found in stack ${stack}. Is the cluster deployed with split DB instances?"
+  else
+    _error "Unknown postgres type: $type"
+  fi
+}
+
+main "$@"