/api/v2/X part 5 (#14718)

* query timestamps are now pre-determined and alignment on timestamps is guarranteed

* turn internal_fatal() to internal_error() to investigate the issue

* handle query when no data exist in the db

* check for non NULL dict when running dictionary garbage collect

* support API v2 requests via ACLK

* add nodes detailed information to /api/v2/nodes

* fixed keys and added dummy nodes for completeness

* added nodes_hard_hash, alerts_hard_hash, alerts_soft_hash; started building a nodes status object to reflect the current status of a node

* make sure replication does not double count charts that are already being replicated

* expose min and max in sts structures

* added view_minimum_value and view_maximum_value; percentage calculation is now an additional pass on the data, removed from formatters; absolute value calculation is now done at the query level, removed from formatters

* respect trimming in percentage calculation; updated swagger

* api/v2/weights preparative work to support multi-node queries - still single node though

* multi-node /api/v2/weights endpoint, supporting all the filtering parameters of /api/v2/data

* when passing the raw option, the query exposes the hidden dimensions

* fix compilation issues on older systems

* the query engine now calculates per dimension min, max, sum, count, anomaly count

* use the macro to calculate storage point anomaly rate

* weights endpoint exposing version hashes

* weights method=value shows min, max, average, sum, count, anomaly count, anomaly rate

* query: expose RESET flag; do not add the same point multiple times to the aggregated point

* weights: more compact output

* weights requests can be interrupted

* all /api/v2 requests can be interrupted and timeout

* allow relative timestamps in weights

* fix macos compilation warnings

* Revert "fix macos compilation warnings"

This reverts commit 8a1d24e41e9b58de566ac59f0c4b1c465bcc0592.

* /api/v2/data group-by now works on dimension names, not ids

* /api/v2/weights does not query metrics without retention and new output format

* /api/v2/weights value and anomaly queries do context queries when contexts are filtered; query timeout is now always in ms

1 files changed, 2 insertions, 0 deletions

diff --git a/health/health_log.c b/health/health_log.c
index d3417493b8..b1f59a1a54 100644
--- a/health/health_log.c
+++ b/health/health_log.c
@@ -95,6 +95,8 @@ inline void health_alarm_log_add_entry(
 ) {
     debug(D_HEALTH, "Health adding alarm log entry with id: %u", ae->unique_id);
 
+    __atomic_add_fetch(&host->health_transitions, 1, __ATOMIC_RELAXED);
+
     // link it
     netdata_rwlock_wrlock(&host->health_log.alarm_log_rwlock);
     ae->next = host->health_log.alarms;