FEATURE: vmauth: allow buffering request bodies before proxying them to backends. This reduces load on backends when processing requests from slow clients such as IoT devices connected to vmauth via slow networks. See #10309 and request body buffering docs.

FEATURE: vmbackupmanager: allow completely disabling scheduled backups by using -disableScheduledBackups command-line flag. This is useful to run vmbackupmanager only for on-demand backups and restores triggered via API. See #10364.

FEATURE: monitoring: take into account all the generated log messages at vm_log_messages_total metric, including suppressed logs if the -loggerLevel command-line flag is set to values other than INFO. Add is_printed label to the vm_log_messages_total metric in order to understand whether the log at the given code location was suppressed or not. This simplifies troubleshooting of VictoriaMetrics components when logs aren't available. See #10304.

FEATURE: vmagent: support configuring different -remoteWrite.queues per remoteWrite url. This allows setting -remoteWrite.queues=1 for backends that do not support out-of-order ingestion (e.g. Mimir), while keeping higher queue counts for other backends such as VictoriaMetrics. Previously, this required running multiple vmagent instances with different queue settings. See #10270.

FEATURE: alerts: add new alerting rules PersistentQueueRunsOutOfSpaceIn12Hours and PersistentQueueRunsOutOfSpaceIn4Hours for vmagent persistent queue capacity. These alerts help users to take proactive actions before vmagent starts dropping metrics due to insufficient persistent queue space. See #10193

FEATURE: All VictoriaMetrics components: add build version information to the home page for consistency with other projects. See #10249.

FEATURE: all VictoriaMetrics components: add flag fs.disableMincore, which allows to disable mincore syscall. See #10327.

FEATURE: vmsingle and vmselect in VictoriaMetrics cluster: expose topN average memory bytes consumption queries in /api/v1/status/top_queries. It can help users to find queries that consume a lot of memory and potentially cause OOM. See #9330.

FEATURE: vmctl: add metrics for tracking the migration progress. See vmctl - monitoring the migration process and #10276.

BUGFIX: vmauth: stop backend health checks for URL prefixes defined in url_map during configuration reloads. Previously, stale backends kept being health-checked and produced repeated warning logs after reloads. See #10334.

BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: properly return /api/v1/status/tsdb response for time range outside partition index. See #10315.

BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: properly report *_requests_total, *_misses_total, *_resets_total, *_syncs_total, *_rotations_total metrics for indexdb/tagFiltersToMetricIDs, indexdb/metricID and indexdb/date_metricID caches. See #10275.

BUGFIX: vmstorage in VictoriaMetrics cluster: correctly return tenants results for /admin/tenants when start or end are specified. See #10312. Thanks to @Defined2014 for the contribution.

BUGFIX: vmui: fix "Percentage from total" calculation on the Cardinality Explorer page when multiple metrics match the filter. See #10323. Thanks to @PleasingFungus for the contribution.

BUGFIX: vmagent: apply -promscrape.maxScrapeSize check to decompressed data instead of compressed data. See #9481.

BUGFIX: vmalert: disallow setting the -notifier.url command-line flag to a null value. See #10355.

BUGFIX: vmalert: fix alert restore when a group contains many rules and is slow to complete evaluation. Previously, the restore process might not retrieve the correct previous alert state. See #10335.

BUGFIX: vmalert: do not skip sending alert notifications to -notifier.url if remote write requests to -remoteWrite.url fail. See #10376.

BUGFIX: MetricsQL: fix changes() function when gaps between samples exceed the lookbehind window. Previously, it could yield a non-zero value even when the sample value remained unchanged. See #10280.

BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: previously ingestion could hit lock contention that triggered frequent context switches and storage connection saturation spikes; now the contention is removed to keep ingestion steady. See #10367.