agent alert notifications redirect (#15350)

* agent alert notifications redirect

* set the same cookies with SameSite: Strict

* registry search now requires only "for" parameter

* registry responses are not cacheable

* fix typo and add more error checking

* registry memory when mmap is used

* fix free with aral

2 files changed, 20 insertions, 25 deletions

diff --git a/health/health.c b/health/health.c
index 4ae7e99974..c39677e250 100644
--- a/health/health.c
+++ b/health/health.c
@@ -81,8 +81,9 @@ static bool prepare_command(BUFFER *wb,
                             const char *crit_alarms,
                             const char *classification,
                             const char *edit_command,
-                            const char *machine_guid)
-{
+                            const char *machine_guid,
+                            uuid_t *transition_id
+) {
     char buf[8192];
     size_t n = 8192 - 1;
 
@@ -188,6 +189,12 @@ static bool prepare_command(BUFFER *wb,
         return false;
     buffer_sprintf(wb, " '%s'", buf);
 
+    char tr_id[UUID_STR_LEN];
+    uuid_unparse_lower(*transition_id, tr_id);
+    if (!sanitize_command_argument_string(buf, tr_id, n))
+        return false;
+    buffer_sprintf(wb, " '%s'", buf);
+
     return true;
 }
 
@@ -575,7 +582,8 @@ static inline void health_alarm_execute(RRDHOST *host, ALARM_ENTRY *ae) {
                               buffer_tostring(crit_alarms),
                               ae->classification?ae_classification(ae):"Unknown",
                               edit_command,
-                              host != localhost ? host->machine_guid:"");
+                              host->machine_guid,
+                              &ae->transition_id);
 
     const char *command_to_run = buffer_tostring(wb);
     if (ok) {
diff --git a/health/notifications/alarm-notify.sh.in b/health/notifications/alarm-notify.sh.in
index 1867cf818f..3cff33db93 100755
--- a/health/notifications/alarm-notify.sh.in
+++ b/health/notifications/alarm-notify.sh.in
@@ -3,7 +3,7 @@
 
 # netdata
 # real-time performance and health monitoring, done right!
-# (C) 2017 Costa Tsaousis <costa@tsaousis.gr>
+# (C) 2023 Netdata Inc.
 # SPDX-License-Identifier: GPL-3.0-or-later
 #
 # Script to send alarm notifications for netdata
@@ -246,7 +246,8 @@ else
   total_crit_alarms="${26}"  # List of alarms in critical state
   classification="${27}"     # The class field from .conf files
   edit_command_line="${28}"  # The command to edit the alarm, with the line number
-  child_machine_guid="${29}" # If populated, the notification is sent for a child
+  child_machine_guid="${29}" # the machine_guid of the child
+  transition_id="${30}"      # the transition_id of the alert
 fi
 
 # -----------------------------------------------------------------------------
@@ -2488,31 +2489,17 @@ urlencode "${value_string}" >/dev/null
 url_value_string="${REPLY}"
 
 redirect_params="host=${url_host}&chart=${url_chart}&family=${url_family}&alarm=${url_name}&alarm_unique_id=${unique_id}&alarm_id=${alarm_id}&alarm_event_id=${event_id}&alarm_when=${when}&alarm_status=${status}&alarm_chart=${chart}&alarm_value=${url_value_string}"
-GOTOCLOUD=0
 
-if [ "${NETDATA_REGISTRY_URL}" == "https://registry.my-netdata.io" ]; then
-  if [ -z "${NETDATA_REGISTRY_UNIQUE_ID}" ]; then
-    if [ -f "@registrydir_POST@/netdata.public.unique.id" ]; then
-      NETDATA_REGISTRY_UNIQUE_ID="$(cat "@registrydir_POST@/netdata.public.unique.id")"
-    fi
-  fi
-  if [ -n "${NETDATA_REGISTRY_UNIQUE_ID}" ]; then
-    GOTOCLOUD=1
-  fi
-fi
-
-if [ ${GOTOCLOUD} -eq 0 ]; then
-  goto_url="${NETDATA_REGISTRY_URL}/goto-host-from-alarm.html?${redirect_params}"
-else
-    # Temporarily disable alarm redirection, as the cloud endpoint no longer exists. This functionality will be restored after discussion on #9487. For now, just lead to netdata.cloud
-    # Re-allow alarm redirection, for alarms 2.0, new template
-  if [ -z "${child_machine_guid}" ]; then
-      goto_url="${NETDATA_REGISTRY_CLOUD_BASE_URL}/alarms/redirect?agentId=${NETDATA_REGISTRY_UNIQUE_ID}&${redirect_params}"
+if [ -z "${NETDATA_REGISTRY_UNIQUE_ID}" ]; then
+  if [ -f "@registrydir_POST@/netdata.public.unique.id" ]; then
+    NETDATA_REGISTRY_UNIQUE_ID="$(cat "@registrydir_POST@/netdata.public.unique.id")"
   else
-      goto_url="${NETDATA_REGISTRY_CLOUD_BASE_URL}/alarms/redirect?agentId=${NETDATA_REGISTRY_UNIQUE_ID}&childId=${child_machine_guid}&${redirect_params}"
+    error "failed to identify this agent via its NETDATA_REGISTRY_UNIQUE_ID."
   fi
 fi
 
+goto_url="${NETDATA_REGISTRY_URL}/registry-alert-redirect.html?agent_machine_guid=${NETDATA_REGISTRY_UNIQUE_ID}&host_machine_guid=${child_machine_guid}&transition_id=${transition_id}&${redirect_params}"
+
 # the severity of the alarm
 severity="${status}"