diff --git a/nix/workbench/backend/nomad-job.nix b/nix/workbench/backend/nomad-job.nix
index 64fd73b0aa8..940e5cd8519 100644
--- a/nix/workbench/backend/nomad-job.nix
+++ b/nix/workbench/backend/nomad-job.nix
@@ -262,6 +262,74 @@ let
           unlimited = false;
         };
 
+        # Prevent allocations from being restarted:
+        ###########################################
+        # Nomad Clients periodically heartbeat to Nomad Servers to confirm they
+        # are operating as expected. By default, Nomad Clients which do not
+        # heartbeat in the specified amount of time are considered down and
+        # their allocations are marked as lost (or disconnected if
+        # "max_client_disconnect" is set) and rescheduled.
+        # This means that if not properly configured allocations running on a
+        # client that fails to heartbeat will be marked "lost" and when the
+        # client reconnects, its allocations, which may still be healthy,
+        # restarted because they have been marked "lost"!!!
+        # See:
+        # - https://developer.hashicorp.com/nomad/docs/configuration/server#client-heartbeats
+        # - https://developer.hashicorp.com/nomad/docs/job-specification/group#stop-after-client-disconnect
+        # - https://developer.hashicorp.com/nomad/docs/job-specification/group#max-client-disconnect
+        # We want these allocations to reconnect without a restart.
+        ### Nomad 1.6.X solution:
+        ### Specifies a duration during which a Nomad client will attempt to
+        ### reconnect allocations after it fails to heartbeat in the
+        ### "heartbeat_grace" window. See the example code below for more
+        ### details. This setting cannot be used with
+        ### "stop_after_client_disconnect".
+        ### When "max_client_disconnect" is specified, the Nomad server will
+        ### mark clients that fail to heartbeat as "disconnected" rather than
+        ### "down", and will mark allocations on a disconnected client as
+        ### "unknown" rather than "lost". These allocations may continue to run
+        ### on the disconnected client. Replacement allocations will be
+        ### scheduled according to the allocations' reschedule policy until the
+        ### disconnected client reconnects. Once a disconnected client
+        ### reconnects, Nomad will compare the "unknown" allocations with their
+        ### replacements and keep the one with the best node score. If the
+        ### "max_client_disconnect" duration expires before the client
+        ### reconnects, the allocations will be marked "lost". Clients that
+        ### contain "unknown" allocations will transition to "disconnected"
+        ### rather than "down" until the last "max_client_disconnect" duration
+        ### has expired.
+        ### https://developer.hashicorp.com/nomad/docs/v1.6.x/job-specification/group#max-client-disconnect
+        max_client_disconnect = "999h";
+        ### Nomad 1.7.X solution:
+        ### (TODO blocker issue https://github.com/hashicorp/nomad/issues/19506)
+        ### Defines the reschedule behaviour of an allocation when the node it
+        ### is running on misses heartbeats. When enabled, if the node it is
+        ### running on becomes disconnected or goes down, this allocations won't
+        ### be rescheduled and will show up as unknown until the node comes back
+        ### up or it is manually restarted.
+        ### This behaviour will only modify the reschedule process on the
+        ### server. To modify the allocation behaviour on the client, see
+        ### "stop_after_client_disconnect" below.
+        ### The unknown allocation has to be manually stopped to run it again.
+        ### Setting `max_client_disconnect` and
+        ### `prevent_reschedule_on_lost = true` at the same time requires that
+        ### rescheduling is disabled entirely (what is done above in the
+        ### reschedule stanza).
+        # prevent_reschedule_on_lost = true;
+        ### Specifies a duration after which a Nomad client will stop
+        ### allocations, if it cannot communicate with the servers. By default,
+        ### a client will not stop an allocation until explicitly told to by a
+        ### server. A client that fails to heartbeat to a server within the
+        ### "heartbeat_grace" window and any allocations running on it will be
+        ### marked "lost" and Nomad will schedule replacement allocations. The
+        ### replaced allocations will normally continue to run on the
+        ### non-responsive client. But you may want them to stop instead — for
+        ###  example, allocations requiring exclusive access to an external
+        ### resource. When specified, the Nomad client will stop them after this
+        ### duration. The Nomad client process must be running for this to
+        ### occur. This setting cannot be used with "max_client_disconnect".
+        # stop_after_client_disconnect = "999h";
+
         # Specifies the restart policy for all tasks in this group. If omitted,
         # a default policy exists for each job type, which can be found in the
         # restart stanza documentation.
diff --git a/nix/workbench/backend/nomad.sh b/nix/workbench/backend/nomad.sh
index 43cf4a1e54a..f6dde0f1528 100644
--- a/nix/workbench/backend/nomad.sh
+++ b/nix/workbench/backend/nomad.sh
@@ -1415,17 +1415,17 @@ backend_nomad() {
         then
           if ! wait_kill_em_all "${jobs_array[@]}"
           then
-            # Don't use fatal here, let `start` decide!
             msg "$(red "Failed to start tracer(s)")"
-            return 1
+            backend_nomad stop-nomad-job "${dir}" || msg "$(red "Failed to stop Nomad Job")"
+            fatal "scenario.sh start-tracers failed!"
           else
             for node in ${nodes[*]}
             do
               if ! test -f "${dir}"/tracer/"${node}"/started
               then
-                # Don't use fatal here, let `start` decide!
                 msg "$(red "Tracer for \"${node}\" failed to start!")"
-                return 1
+                backend_nomad stop-nomad-job "${dir}" || msg "$(red "Failed to stop Nomad Job")"
+                fatal "scenario.sh start-tracers failed!"
               fi
             done
           fi
@@ -3487,7 +3487,9 @@ client {
   servers = [ ${servers_addresses} ]
   # Sets the search path that is used for CNI plugin discovery. Multiple paths can
   # be searched using colon delimited paths
-  cni_path = "${cni_plugins_path}"
+# TODO: needed to allow having more than one Nomad profile running locally
+# Nomad 1.7.X fails somewhat silently when reading this configuration option.
+#  cni_path = "${cni_plugins_path}"
   # Specifies the maximum amount of time a job is allowed to wait to exit.
   # Individual jobs may customize their own kill timeout, but it may not exceed
   # this value.
diff --git a/nix/workbench/backend/nomad/exec.nix b/nix/workbench/backend/nomad/exec.nix
index e1fee1b673d..3b658b492db 100644
--- a/nix/workbench/backend/nomad/exec.nix
+++ b/nix/workbench/backend/nomad/exec.nix
@@ -20,22 +20,8 @@ let
   extraShellPkgs = let
     # If we are going to use the `exec` driver we use the SRE patched version of
     # Nomad that allows to use `nix_installables` as artifacts.
-    nomad-sre = (pkgs.buildGo119Module rec {
-      pname = "nomad";
-      version = "1.4.3";
-      subPackages = [ "." ];
-      doCheck = true;
-      src = pkgs.fetchFromGitHub { # "github:input-output-hk/nomad/release/1.4.3"
-        owner = "input-output-hk";
-        repo = pname;
-        rev = "2b8a93390"; # Use to be "release/${version}" but it changes.
-        # nix-prefetch-url --unpack https://github.com/input-output-hk/nomad/archive/2b8a93390/1.4.3.tar.gz
-        sha256 = "0l2sfhpg0p5mjdbipib7q63wlsrczr2fkq9xi641vhgxsjmprvwm";
-      };
-      # error: either `vendorHash` or `vendorSha256` is required
-      # https://discourse.nixos.org/t/buildgomodule-how-to-get-vendorsha256/9317
-      vendorSha256 = "sha256-JQRpsQhq5r/QcgFwtnptmvnjBEhdCFrXFrTKkJioL3A=";
-    });
+    commit = "8f3b74796a8f56f38a812813c64dba995956a66e"; # Patched 1.6.3
+    nomad-sre = (__getFlake "github:input-output-hk/cardano-perf/${commit}").packages.x86_64-linux.nomad;
   in
     [ nomad-sre
       # The HTTP server to upload/download the genesis tar file in a local env.
diff --git a/nix/workbench/backend/nomad/exec.sh b/nix/workbench/backend/nomad/exec.sh
index 26966077694..42b6f9da724 100644
--- a/nix/workbench/backend/nomad/exec.sh
+++ b/nix/workbench/backend/nomad/exec.sh
@@ -186,8 +186,6 @@ deploy-genesis-nomadexec() {
   local usage="USAGE: wb backend $op RUN-DIR"
   local dir=${1:?$usage}; shift
   local nomad_job_name=$(jq -r ". [\"job\"] | keys[0]" "${dir}"/nomad/nomad-job.json)
-  local server_name=$(envjqr 'nomad_server_name')
-  local client_name=$(envjqr 'nomad_client_name')
 
   # Add genesis to HTTP cache server
   local nomad_agents_were_already_running=$(envjqr 'nomad_agents_were_already_running')
@@ -199,8 +197,8 @@ deploy-genesis-nomadexec() {
       if test "${nomad_agents_were_already_running}" = "false"
       then
         msg "$(red "Startup of webfs failed, cleaning up ...")"
+        # `stop-nomad-job` takes care of stopping the Nomad agents.
         backend_nomad stop-nomad-job "${dir}" || msg "$(red "Failed to stop Nomad Job")"
-        wb_nomad agents stop "${server_name}" "${client_name}" "exec"
       fi
       fatal "Failed to start a local HTTP server"
     fi
@@ -213,8 +211,8 @@ deploy-genesis-nomadexec() {
     if test "${nomad_agents_were_already_running}" = "false"
     then
       msg "$(red "Startup of webfs failed, cleaning up ...")"
+      # `stop-nomad-job` takes care of stopping the Nomad agents.
       backend_nomad stop-nomad-job "${dir}" || msg "$(red "Failed to stop Nomad Job")"
-      wb_nomad agents stop "${server_name}" "${client_name}" "exec"
     fi
     fatal "Failed to add genesis file to local HTTP server"
   else
@@ -225,8 +223,8 @@ deploy-genesis-nomadexec() {
   if ! backend_nomad deploy-genesis-wget "${dir}" "${uri}"
   then
     msg "$(red "Deploy of genesis failed, cleaning up ...")"
+    # `stop-nomad-job` takes care of stopping the Nomad agents.
     backend_nomad stop-nomad-job "${dir}" || msg "$(red "Failed to stop Nomad Job")"
-    wb_nomad agents stop "${server_name}" "${client_name}" "exec"
     fatal "Deploy of genesis \"${uri}\" failed"
   else
     msg "$(green "Genesis \"${uri}\" deployed successfully")"
diff --git a/nix/workbench/nomad.sh b/nix/workbench/nomad.sh
index d31f45031ab..b55c92e6607 100644
--- a/nix/workbench/nomad.sh
+++ b/nix/workbench/nomad.sh
@@ -83,6 +83,7 @@ usage_nomad() {
     $(helpcmd job monitor-alloc-id)
     $(helpcmd job monitor-alloc-id-task-name)
     $(helpcmd job task-name-allocation-id)
+    $(helpcmd job task-name-node-name)
 EOF
 }
 
@@ -633,7 +634,7 @@ EOL
           # can represent an abnormal exit / uknown state!
           if wb_nomad server is-running "${name}"
           then
-            msg "$(red "FATAL: Nomad server \"${name}\" is already running or in an uknown state, call 'wb nomad server stop ${name}' or 'wb nomad nuke' first")"
+            msg "$(red "FATAL: Nomad server \"${name}\" is already running or in an unknown state, call 'wb nomad server stop ${name}' or 'wb nomad nuke' first")"
             return 1
           else
             local state_dir=$(wb_nomad server state-dir-path "${name}")
@@ -709,7 +710,7 @@ EOL
           # Checks
           if wb_nomad server is-running "${name}"
           then
-            msg "$(red "FATAL: Nomad server \"${name}\" is already running or in an uknown state, call 'wb nomad server stop ${name}' or 'wb nomad nuke' first")"
+            msg "$(red "FATAL: Nomad server \"${name}\" is already running or in an unknown state, call 'wb nomad server stop ${name}' or 'wb nomad nuke' first")"
             return 1
           fi
           # Start `nomad` server".
@@ -741,7 +742,7 @@ EOL
               msg "$(yellow "port \"127.0.0.1:${http_port}\" not ready")"
               msg "$(yellow "Check logs (${state_dir})")"
               # Let the "stop" subcommand clean everything!
-              wb_nomad server stop "${name}"
+              wb_nomad server stop "${name}" || true
               return 1
             fi
             echo -ne "\b\b\b"
@@ -755,30 +756,36 @@ EOL
           local name=${1:?$usage}; shift
           # Stop Nomad server by name
           local pids=$(wb_nomad server pids-array "${name}")
-          for pid_number in ${pids[@]}
-          do
-            msg "$(blue Stopping) Nomad $(yellow "server \"${name}\"") process PID ${pid_number} ..."
-            if ! kill -SIGINT "${pid_number}" >/dev/null 2>&1
-            then
-              msg "$(red "Killing PID ${pid_number} failed")"
-            else
-              # Wait 15 seconds for the process to fully exit or kill it.
-              msg "$(blue Wait) up to 15 seconds for PID ${pid_number} to exit"
-              timeout 15 tail --pid="${pid_number}" -f /dev/null || true
-              if kill -0 "${pid_number}" >/dev/null 2>&1
+          if test -z "${pids}"
+          then
+            msg "$(red "Found no running Nomad server process to stop, manually clean possible remaining processes")"
+            return 1
+          else
+            for pid_number in ${pids[@]}
+            do
+              msg "$(blue Stopping) Nomad $(yellow "server \"${name}\"") process PID ${pid_number} ..."
+              if ! kill -SIGINT "${pid_number}" >/dev/null 2>&1
               then
-                msg "$(yellow "Timeout killing PID ${pid_number}, trying SIGKILL")"
-                kill -SIGKILL "${pid_number}" >/dev/null 2>&1 || true
+                msg "$(red "Killing PID ${pid_number} failed")"
+              else
+                # Wait 15 seconds for the process to fully exit or kill it.
+                msg "$(blue Wait) up to 15 seconds for PID ${pid_number} to exit"
+                timeout 15 tail --pid="${pid_number}" -f /dev/null || true
+                if kill -0 "${pid_number}" >/dev/null 2>&1
+                then
+                  msg "$(yellow "Timeout killing PID ${pid_number}, trying SIGKILL")"
+                  kill -SIGKILL "${pid_number}" >/dev/null 2>&1 || true
+                fi
               fi
-            fi
-          done
-          # Remove PID file if process was really killed (or wasn't running)!
-          if test -z "$(wb_nomad server pids-array "${name}")"
-          then
-            local pid_file=$(wb_nomad server pid-filepath "${name}")
-            if test -f "${pid_file}"
+            done
+            # Remove PID file if all processes were killed!
+            if test -z "$(wb_nomad server pids-array "${name}")"
             then
-              rm "${pid_file}"
+              local pid_file=$(wb_nomad server pid-filepath "${name}")
+              if test -f "${pid_file}"
+              then
+                rm "${pid_file}"
+              fi
             fi
           fi
         ;;
@@ -997,7 +1004,7 @@ EOL
               msg "$(yellow "port \"127.0.0.1:${http_port}\" not ready")"
               msg "$(yellow "Check logs (${state_dir})")"
               # Let the "stop" subcommand clean everything!
-              wb_nomad client stop "${name}"
+              wb_nomad client stop "${name}" || true
               return 1
             fi
             echo -ne "\b\b\b"
@@ -1017,7 +1024,7 @@ EOL
               msg "$(yellow "Nomad client not connected to Nomad server")"
               msg "$(yellow "Check logs (${state_dir})")"
               # Let the "stop" subcommand clean everything!
-              wb_nomad client stop "${name}"
+              wb_nomad client stop "${name}" || true
               return 1
             fi
             echo -ne "\b\b\b"
@@ -1079,42 +1086,45 @@ EOL
           fi
           # Stop Nomad client by name
           local pids=$(wb_nomad client pids-array "${name}")
-          for pid_number in ${pids[@]}
-          do
-            msg "$(blue Stopping) Nomad $(yellow "client \"${name}\"") process PID ${pid_number} ..."
-            local cmd_array=("${root_prefix}" "bash" "-c")
-            if ! ${cmd_array[@]} "kill -SIGINT ${pid_number}" >/dev/null 2>&1
-            then
-              msg "Killing PID ${pid_number} failed"
-            else
-              # Wait 15 seconds for the process to fully exit or kill it.
-              msg "$(blue Wait) up to 30 seconds for PID ${pid_number} to exit"
-              timeout 30 tail --pid="${pid_number}" -f /dev/null || true
+          if test -z "${pids}"
+          then
+            msg "$(red "Found no running Nomad client process to stop, manually clean possible remaining processes")"
+            return 1
+          else
+            for pid_number in ${pids[@]}
+            do
+              msg "$(blue Stopping) Nomad $(yellow "client \"${name}\"") process PID ${pid_number} ..."
               local cmd_array=("${root_prefix}" "bash" "-c")
-              if ${cmd_array[@]} "kill -0 ${pid_number}" >/dev/null 2>&1
+              if ! ${cmd_array[@]} "kill -SIGINT ${pid_number}" >/dev/null 2>&1
               then
-                msg "$(yellow "Timeout killing PID ${pid_number}, trying SIGKILL")"
+                msg "Killing PID ${pid_number} failed"
+              else
+                # Wait 15 seconds for the process to fully exit or kill it.
+                msg "$(blue Wait) up to 30 seconds for PID ${pid_number} to exit"
+                timeout 30 tail --pid="${pid_number}" -f /dev/null || true
                 local cmd_array=("${root_prefix}" "bash" "-c")
-                ${cmd_array[@]} "kill -SIGKILL ${pid_number}" >/dev/null 2>&1 || true
+                if ${cmd_array[@]} "kill -0 ${pid_number}" >/dev/null 2>&1
+                then
+                  msg "$(yellow "Timeout killing PID ${pid_number}, trying SIGKILL")"
+                  local cmd_array=("${root_prefix}" "bash" "-c")
+                  ${cmd_array[@]} "kill -SIGKILL ${pid_number}" >/dev/null 2>&1 || true
+                fi
               fi
-            fi
-          done
-          # Remove PID file if process was really killed (or wasn't running)!
-          if test -z "$(wb_nomad client pids-array "${name}")"
-          then
-            # WHY? The client is keeping some directories mounted!
-            # Maybe because of the 2 processes it creates (testes running
-            # only one client instance), I may be killing a child first?
-            # Or the timeout needs more time?
-            msg "Unmount any folders left by the client"
-            local cmd_array=("${root_prefix}" "bash" "-c")
-            # Command fails when there's nothing to umount!
-            grep "${state_dir}" /proc/mounts | cut -f2 -d" " | sort -r | ${cmd_array[@]} 'xargs -I "{}" umount -n "{}"' || true
-            # Now mark as "not running"
-            local pid_file=$(wb_nomad client pid-filepath "${name}")
-            if test -f "${pid_file}"
+            done
+            # Remove PID file if all processes were killed!
+            if test -z "$(wb_nomad client pids-array "${name}")"
             then
-              rm "${pid_file}"
+              # WHY? The client is keeping some directories mounted!
+              msg "Unmount any folders left by the client"
+              local cmd_array=("${root_prefix}" "bash" "-c")
+              # Command fails when there's nothing to umount!
+              grep "${state_dir}" /proc/mounts | cut -f2 -d" " | sort -r | ${cmd_array[@]} 'xargs -I "{}" umount -n "{}"' || true
+              # Now mark as "not running"
+              local pid_file=$(wb_nomad client pid-filepath "${name}")
+              if test -f "${pid_file}"
+              then
+                rm "${pid_file}"
+              fi
             fi
           fi
         ;;
@@ -1405,10 +1415,34 @@ EOF
       local nomad_clients_dir="$(wb_nomad dir-path client)"
       # Nuke all Nomad clients
       for client_name in $(ls "${nomad_clients_dir}"); do
+        msg "Config folder of Nomad client \"${client_name}\" found"
         if wb_nomad client is-running "${client_name}"
         then
-          wb_nomad client stop "${client_name}"
-          wb_nomad client cleanup "${client_name}"
+          msg "Nomad client \"${client_name}\" is running"
+          if wb_nomad client stop "${client_name}"
+          then
+            # Only call cleanup if stop did not fail
+            wb_nomad client cleanup "${client_name}"
+          else
+            msg "Failed to stop Nomad client \"${client_name}\", now in unknown state, manual cleanup of processes needed"
+          fi
+        else
+          msg "Nomad client \"${client_name}\" is not running"
+        fi
+        # Nuke the client's dir
+        local state_dir=$(wb_nomad client state-dir-path "${client_name}")
+        msg "Removing \"${state_dir}\" ..."
+        local root_prefix
+        if test -e "${state_dir}"/root
+        then
+          root_prefix=$(cat "${state_dir}"/root)
+        else
+          root_prefix=""
+        fi
+        local cmd_array=("${root_prefix}" "bash" "-c")
+        if ! ${cmd_array[@]} "rm -rf ${state_dir}" >/dev/null 2>&1
+        then
+          msg "Failed to remove config folder of Nomad client \"${client_name}\", now in unknown state, manual cleanup needed"
         fi
       done
       # Nuke the nomad-driver-podman plugin
@@ -1423,16 +1457,32 @@ EOF
       fi
       # Nuke all Nomad servers
       for server_name in $(ls "${nomad_servers_dir}"); do
+        msg "Config folder of Nomad server \"${server_name}\" found"
         if wb_nomad server is-running "${server_name}"
         then
-          wb_nomad server stop "${server_name}"
-          wb_nomad server cleanup "${server_name}"
+          msg "Nomad server \"${server_name}\" is running"
+          if wb_nomad server stop "${server_name}"
+          then
+            # Only call cleanup if stop did not fail
+            wb_nomad server cleanup "${server_name}"
+          else
+            msg "Failed to stop Nomad server \"${server_name}\", now in unknown state, manual cleanup of processes needed"
+          fi
+        else
+          msg "Nomad server \"${server_name}\" is not running"
+        fi
+        # Nuke the server's dir
+        local state_dir=$(wb_nomad server state-dir-path "${server_name}")
+        msg "Removing \"${state_dir}\" ..."
+        if ! rm -rf "${state_dir}" >/dev/null 2>&1
+        then
+          msg "Failed to remove config folder of Nomad server \"${server_name}\", now in unknown state, manual cleanup needed"
         fi
       done
       # Nuke the Nomad Agents' .cache dir
       # Keep top level Nomad cache dir because it includes Vault's dirs.
-      rm -rf "${nomad_servers_dir}" >/dev/null 2>&1
       rm -rf "${nomad_clients_dir}" >/dev/null 2>&1
+      rm -rf "${nomad_servers_dir}" >/dev/null 2>&1
       # Bye HTTP server
       if wb_nomad webfs is-running
       then
@@ -2105,6 +2155,13 @@ EOF
           local task_name=${1:?$usage}; shift
           jq -r '.ID' "${job_file}".run/task.${task_name}.final.json
         ;;
+####### job -> task-name-node-name )############################################
+        task-name-node-name )
+          local usage="USAGE:wb nomad ${op} ${subop} JOB-FILE TASK-NAME"
+          local job_file=${1:?$usage};  shift
+          local task_name=${1:?$usage}; shift
+          jq -r '.NodeName' "${job_file}".run/task.${task_name}.final.json
+        ;;
 ####### job -> stop )###########################################################
         stop )
           local usage="USAGE: wb nomad ${op} ${subop} JOB-FILE TASK-NAME"
diff --git a/nix/workbench/scenario.sh b/nix/workbench/scenario.sh
index 90744743511..5bb68f4b965 100644
--- a/nix/workbench/scenario.sh
+++ b/nix/workbench/scenario.sh
@@ -31,7 +31,16 @@ fi
 case "$op" in
     idle )
         backend start-tracers        "$dir"
+
+        scenario_setup_exit_trap              "$dir"
+        # Trap start
+        ############
         backend start-nodes          "$dir"
+        # Trap end
+        ##########
+        scenario_cleanup_termination
+
+        backend stop-all             "$dir"
         ;;
 
     tracer-only )