v1.129.1

Compare Source

v1.129.1

Released at 2025-11-04

• BUGFIX: vminsert, vmagent and vmsingle: properly apply maxDataSize memory limits to the snappy encoded requests. It protects ingest endpoints from malicious requests.

v1.129.0

Compare Source

v1.129.0

Released at 2025-10-31

• FEATURE: vminsert and vmstorage in VictoriaMetrics cluster: introduce new RPC protocol for insert-storage communication. See this PR #​9820 for details.

• FEATURE: vmalert: explicitly check response type for range queries during replay and return error on type mismatch. This change should reduce confusions like in #​9779.

• FEATURE: vmctl: allow providing multiple filters for remote-read migration mode via multiple --remote-read-filter-label and --remote-read-filter-label-value flags. This is useful in order to narrow down the data being migrated by using more precise filters. See this PR #​9917 for details.

• FEATURE: vmalert: support alert_relabel_configs per each notifier in -notifier.config file. See #​5980.

• FEATURE: vmalert: add template function now to return the Unix timestamp in seconds at the time of the template evaluation. For example, {{ (now | toTime).Sub $activeAt }} can calculate duration the alert has been active. See this issue #​9864 for more details. Thank you @​nguu0123 for the pull request.

• FEATURE: vmalert: start groups with group.interval > 5m faster. Before, vmalert was delaying group start for random duration up to group.interval. This could significantly delay group's evaluation if their interval was set to big values. With this change, vmalert limits the max start delay via new command-line flag --group.maxStartDelay (by default, 5m).

• FEATURE: vmagent: add /remotewrite-relabel-config and /api/v1/status/remotewrite-relabel-config API handlers for returning content of --remoteWrite.relabelConfig cmd-line flag. Useful for inspecting and verifying the final global relabeling rules used by vmagent. Access to new handlers can be protected via --configAuthKey command-line flag. See #​9504.

• FEATURE: vmagent: add /remotewrite-url-relabel-config and /api/v1/status/remotewrite-url-relabel-config API handlers for returning content of --remoteWrite.urlRelabelConfig cmd-line flag. Useful for inspecting and verifying the final per-remote-write relabeling rules used by vmagent. Access to new handlers can be protected via --configAuthKey command-line flag. See #​9504.

• FEATURE: vmagent: improve performance for the stream aggregation with multiple configured rules. See this issue #​9878 for details.

• FEATURE: stream aggregation: change the behavior when both streamAggr.dropInput and streamAggr.keepInput are set to true. Previously, all input samples were kept when both flags were true. Now, only input samples matching any aggregation are retained; all others are dropped. See #​9724.

• FEATURE: VictoriaMetrics enterprise: slow query statistics logging is now enabled by default with -search.logSlowQueryStats=5s. This makes it easier to analyze query performance via the Query Stats dashboard. See Query execution stats for more details.

• FEATURE: vmauth: make the load distribution more even among the backends which execute queries with various durations. See #​9712.

• FEATURE: dashboards/all: enable column filters in Non-default flags panel. See #​9910. Thanks to @​SamarthBagga for enhancement!

• FEATURE: add linux/s390x artifact to releases. See #​9697 for the details.

• BUGFIX: vmbackup, vmrestore, vmbackupmanager: complete a fix of environment variables configuration parsing for connection to AWS S3. Previously, such settings were ignored starting from v1.115.0 and releases v1.128.0, v1.122.6 and v1.110.21 did not fix an issue completely. See this issue #​9858 for details.

• BUGFIX: vmalert: fix search over group names and other attributes in vmalert's WEB UI. This functionality was broken since v1.117.0. See #​9886 for details.

• BUGFIX: vmalert: preserve HTML formatting in rule's annotations in vmalert's WEB UI and vmui's alerting page. See #​9892 for details.

• BUGFIX vmbackupmanager: consistently set vm_backup_last_created_at{type="latest"} to the latest backup start time.

• BUGFIX: vmbackupmanager: prevent breaking shell formatting when using vmbackupmanager CLI mode by adding a newline after the command output. Previously, using CLI mode would not always include newline at the end of output, which could break shell formatting.

v1.128.0

Compare Source

v1.128.0

Released at 2025-10-17

Update Note 1: vmsingle and vmagent: default value of the flag -promscrape.dropOriginalLabels changed from false to true.
It disables Discovered targets debug IU by default.

• SECURITY: upgrade Go builder from Go1.25.1 to Go1.25.3. See the list of issues addressed in Go1.25.3.

• SECURITY: upgrade base docker image (Alpine) from 3.22.1 to 3.22.2. See Alpine 3.22.2 release notes.

• FEATURE: vmagent: add opentelemetry format for kafka consumer. See this issue #​9734 for details.

• FEATURE: vmagent and vmsingle: add -opentelemetry.convertMetricNamesToPrometheus command-line flag, which can be used for enabling automatic conversion of the ingested metric names into Prometheus-compatible format. See these docs and this issue #​9830.

• FEATURE: vmagent, vmalert: add -secret.flags command-line flag to configure flags to be hidden in logs and on /metrics. This is useful for protecting sensitive flag values (for example -remoteWrite.headers) from being exposed in logs or metrics. See #​6938. Thank you @​truepele for the issue and PR

• FEATURE: vmagent: change default value of the flag -promscrape.dropOriginalLabels from false to true. This helps reducing CPU and Memory usage by dropping targets original labels during service discovery, but disables relabel debugging UI. This change gives constant improvement on resource usage, and relabel debugging can be turned on on demand. See this issue #​9665 for details.

• FEATURE: vmui: on the Cardinality Explorer page, rename Reset to Reset filters and disable the button when no filters are modified. See #​9609.

• FEATURE: vmui: reflect query visibility state in URL on Query or Raw Query pages. Now, when sharing URL with a query, its visibility state will be preserved.

• BUGFIX: vmui: fix issue where updating one query parameter removed others. See #​9816 for details.

• BUGFIX: vmui alerting tab: update the search filter if the corresponding query argument was passed. Reset selected items in dropdown filters if they are not available anymore after the update.

• BUGFIX: vmbackup, vmrestore, vmbackupmanager: properly apply additional configuration of connection provided via environment variables (such as AWS_CA_BUNDLE). Previously, such settings were ignored starting from v1.115.0. See this issue #​9858 for details.

• BUGFIX: dashboards/victoriametrics, dashboards/victoriametrics-cluster: fixed full disk space ETA queries, before average size of a row in indexdb was one for all vm storages, now it's calculated per each storage.

• BUGFIX: stream aggregation: remove impact of flush_on_shutdown on timestamps of flushed samples. See #​9642.

v1.127.0

Compare Source

v1.127.0

Released at 2025-10-03

• SECURITY: upgrade Go builder from Go1.25.0 to Go1.25.1. See the list of issues addressed in Go1.25.1.

• FEATURE: vmalert: add -rule.resultsLimit command-line flag to allow limiting the number of alerts or recording results a single rule can produce. See #​5792.

• FEATURE: vmauth: stream responses from backends to clients without delays. Previously the backend data could be buffered at vmauth side for indefinite amounts of time. This was preventing from using vmauth for streaming the data from backends in live tailing mode. See VictoriaLogs#667.

• FEATURE: vmbackup, vmrestore: push metrics to configured -pushmetrics.url on shutdown. Before, if -pushmetrics.url was configured, vmbackup or vmrestore might have skipped to report their metrics before shutdown.

• FEATURE: dashboards/operator, dashboards/query-stats: add ad-hoc filters.

• BUGFIX: vmagent: remove the error log when marshaling an invalid comment or an empty HELP metadata line during scraping, if metadata processing is enabled. See #​9710.

• BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: prevent unexpected performance degradation caused by cache misses (exposed via vm_cache_misses_total metric) during rotation. See this PR #​9769 for details.

• BUGFIX: all components: restore sorting order of summary and quantile metrics exposed by VictoriaMetrics components on /metrics page. See metrics#105 for details.

• BUGFIX: vmsingle and vmselect in VictoriaMetrics cluster: avoid applying offset modifier twice to the request time when an instant query uses rollup functions rate() or avg_over_time() with cache enabled. See #​9762.

• BUGFIX: vmalert: restore support for query templates in alert rule labels after the regression introduced in #​9543. See #​9783 for details.

v1.126.0

Compare Source

v1.126.0

Released at 2025-09-12

• FEATURE: vmagent: respect enable.auto.commit setting for kafka consumer configuration.

• FEATURE: vmauth: add an ability to merge client query args with the query args specified at backend urls. This allows merging extra_filters args specified at the particular VictoriaLogs backend in vmauth config and the extra_filters args specified in the Grafana datasource for VictoriaLogs. This is needed for VictoriaLogs#106. See VMAuth - Query args handling docs for more details.

• FEATURE: dashboards/victoriametrics-cluster: add panel Storage full ETA in the vmstorage section to display the minimum approximate time across all the nodes to reach 100% of allowed disk capacity.

• FEATURE: vmui: show vmselect version in the vmui's footer.

• BUGFIX: vmalert: fix possible partial rule update responses in group-related APIs during group updates in hot config reload. See #​9551

• BUGFIX: vmui: properly apply rollup functions to metrics based on their name in vmui's metrics explorer. See #​9655 for details. Thanks to @​wbwren-eric for the fix.

• BUGFIX: all VictoriaMetrics enterprise components: fix support for automatic issuing of TLS certificates for HTTPS server via Let's Encrypt service using TLS-ALPN-01 challenge. See Automatic issuing of TLS certificates for more info.

• BUGFIX: vmui: fix VMUI backend URL, while using multitenant API. See more in #​9703.

• BUGFIX: all components: properly expose metadata for summaries and histograms in VictoriaMetrics components with enabled -metrics.exposeMetadata cmd-line flag. See metrics#98 for details.

• BUGFIX: all components: lower severity of the log message for unavailable Pressure Stall Information (PSI) metrics from ERROR to INFO level. See #​9161 for details.

• BUGFIX: vmbackupmanager: properly prepare restore mark contents when using a backup shortname (e.g. vmbackupmanager restore create daily/2025-09-12). Previously, restore would fail with failed to restore backup: cannot initialize remote fs: missing scheme in path error.

v1.125.1

Compare Source

v1.125.1

Released at 2025-09-03

• BUGFIX: vmselect in VictoriaMetrics cluster: properly route requests for prometheus/vmui/config.json API. Follow-up after 7f15e9f.
• BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: fix workingsetcache metrics. Previously, after cache rotation, metrics could be double-counted or inflated. See 9553 for details.
• BUGFIX: vmui: fix the issue where filtering on click does not work on the Explorer Cardinality page. See #​9674 for details.

v1.125.0

Compare Source

v1.125.0

Released at 2025-08-29

• FEATURE: vmsingle and vmselect in VictoriaMetrics cluster: apply -search.maxQueryLen limit to Graphite queries. Previously, this limit was only applied to Prometheus queries.

• FEATURE: upgrade Go builder from Go1.24.6 to Go1.25. See Go1.25 release notes.

• FEATURE: vmui: add export functionality for Query (Table view) and RawQuery tabs in CSV/JSON format. See #​9332.

• FEATURE: vmui: replace Alerts tab with Alerting tab in vmui. The new Alerting tab displays vmalert groups and rules directly in vmui interface without redirecting user to vmalert's WEB UI. Links of format .*/prometheus/vmalert/.* will continue working by redirecting to vmalert's UI. This functionality is available only if -vmalert.proxyURL is set on vmselect. Some functionality of the new Alerting tab requires vmalert to be of the same version as vmselect, or higher.

• FEATURE: vmalert: add /api/v1/group?group_id=<id> API endpoint for viewing details of a specific rules group. The new handler is used by new Alerting tab in vmselect.

• FEATURE: vmalert: add lastError field to /api/v1/notifiers response. The new field contains an error message if the last attempt to send data to the notifier failed. The error can be also viewed in Alerting. Notifiers tab.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: optimize /api/v1/labels and /api/v1/label/TAG/values requests with a single match or extra_filters filter containing timeseries metric name. See this PR #​9489 for details.

• BUGFIX: vmagent: prevent remote write ingestion stop on push error for Google Pub/Sub integration.

• BUGFIX: vmauth: properly handle mTLS authorization and routing. Previously it didn't work. See #​29.

• BUGFIX: MetricsQL: fix timestamp function compatibility with Prometheus when used with sub-expressions such as timestamp(sum(foo)). The fix applies only when -search.disableImplicitConversion flag is set. See more in #​9527-comment and metricsql#55.

• BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: optimize subtract operation on uint64 sets. This should potentially improve index search with huge number of deleted series. See this issue 9602 for details.

• BUGFIX: all VictoriaMetrics enterprise components: fix support for automatic issuing of TLS certificates for HTTPS server at -httpListenAddr via Let's Encrypt service. See these docs.

v1.124.0

Compare Source

v1.124.0

Released at 2025-08-15

Update Note 1: vmsingle and vmstorage in VictoriaMetrics cluster: performance regression for queries that match previously deleted time series. The issue affects installation that previously deleted big number of time series but continue querying them. More details in #​9602. The degradation will be addressed in upcoming releases.

• SECURITY: upgrade Go builder from Go1.24.5 to Go1.24.6. See the list of issues addressed in Go1.24.6.

• FEATURE: vmsingle and vmselect in VictoriaMetrics cluster: protect graphite /render API endpoint with new flag -search.maxGraphitePathExpressionLen. See this PR #​9534 for details.

• FEATURE: expose vm_total_disk_space_bytes metric at the /metrics page, which shows the total disk space for the data directory specified via -storageDataPath. This metric can be useful for building alerts and graphs for the percentatge of free disk space via vm_free_disk_space_bytes / vm_total_disk_space_bytes. See this comment.

• FEATURE: all: leave non-existing environment variables as is in config files instead of failure. For example, if the file referred by -promscrape.config contains %{NON_EXISTING_ENV_VAR} placeholder, then it is left as is instead of failing to load the file. This simplifies the usage of environment variables in config files and in command-line flags according to these docs. Users can easily notice non-existing env vars in config files and in command-line flags by looking at their values - they will literally contain %{NON_EXISTING_ENV_VAR} strings.

• FEATURE: vmselect, vminsert and vmstorage in VictoriaMetrics cluster: add rpc handshake timeout configuration via -rpc.handshakeTimeout flag (default 5s). Set deadline for the entire handshake process instead of per-operation timeout. See #​9345 for more details.

• FEATURE: vmagent: add -enableMetadata command-line flag to allow sending metadata to the configured -remoteWrite.url, metadata can be scraped from targets, received via VictoriaMetrics remote write, Prometheus remote write v1 or OpenTelemetry protocol. See #​2974.

• BUGFIX: vmagent: treat single remoteWrite.disableOnDiskQueue flag the same way as if it is explicitly specified for each remote write url. See #​9565 for details.

• BUGFIX: vmalert-tool: print a proper error message when templating function fails during execution. Previously, vmalert-tool could throw a misleading panic message instead.

• BUGFIX: vmauth: properly read proxy-protocol header. See this PR #​9546 for details.

• BUGFIX: dashboards/vmagent: fix samples rate panel not showing data in case vmagent is not scraping metrics. Previously, the panel would not display "samples in" results if vmagent only accepts metrics via push protocols.

• BUGFIX: vmselect in VictoriaMetrics cluster: prevent resource leak on partial responses for /federate API requests. This this PR 9536 for details. Thanks to the @​fxrlv.

• BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: prevent performance degradation on hitting daily or hourly series cardinality limits. See this issue 9554 for details.

• BUGFIX: MetricsQL: gracefully handle histogram_quantile calculation for histograms where buckets are created on demand. This change isn't needed for users who use native instrumentation SDKs for metrics from VictoriaMetrics, Prometheus or OpenTelemetry.

• BUGFIX: MetricsQL: return a proper error message when the function argument is expected to be a string or scalar.

• BUGFIX: dashboards/victoriametrics-cluster: fix panels showing 99th percentile of series or samples read per query or per series. Before, panels were showing the summarized value across all vmselect instances, which didn't make much sense. Now, panels show the max value across the vmselect instances, making it easier to understand complexity of the heaviest queries served.

• BUGFIX: vmalert: fix potential data race and missing firing states when replaying alerting rule with -replay.ruleEvaluationConcurrency>1.

• BUGFIX: vmalert: fix the {{ $activeAt }} variable value in annotation templating when the alert has already triggered. See this issue #​9543 for details.

• BUGFIX: vmbackup, vmbackupmanager: allow enabling checksum calculation for PUT requests by using -s3ChecksumAlgorithm command-line flag. This is required for S3 configurations with WORM being enabled. See #​9532.

v1.123.0

Compare Source

v1.123.0

Released at 2025-08-01

• FEATURE: vmsingle and vmagent: add ipv6 address as a metadata label to gce_sd_configs. See #​9370 PR. Thanks to the @​dstevensson.

• FEATURE: all the VictoriaMetrics components: support NetBSD builds. See this #​9473 PR for details. Thanks to the @​iamleot.

• FEATURE: vmselect in VictoriaMetrics cluster: disable retries for requests to vmstorage when handshake or dial errors occur. Retries in such cases are ineffective and can amplify the problem. See #​9345 and #​9484 for details.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: introduce -storage.idbPrefillStart flag (default: 1h) to control how early next indexdb prefill starts before rotation. Increasing the value may reduce CPU spikes during rotation. See #​9393-comment for details.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: Add support for dynamically enabling new series logging via the /internal/log_new_series API and corresponding authFlag -logNewSeriesAuthKey. The logging is enabled for one minute and is automatically disabled afterward. This is a more convenient alternative to the -logNewSeries flag. See 8879 for details. Thanks to the @​leiwingqueen

• FEATURE: vmagent and vmsingle: improve Kubernetes service discovery performance with large amount of configured role selectors. See this #​9354 issue for details. Thanks to the @​fxrlv

• FEATURE: vmctl: add an option to change path for temporary files storage when migrating from Prometheus snapshots. See #​9505 for the details.

• FEATURE: vmagent and vmsingle: improve Kubernetes service discovery performance with large amount of configured role selectors. See this #​9354 issue for details. Thanks to the @​fxrlv

• BUGFIX: vmauth: do not configure -httpListenAddr.useProxyProtocol for -httpInternalListenAddr. See this issue #​9515 for details.

• BUGFIX: vmui: always display the tenant selector if the list of tenants is not empty. See #​9396.

v1.122.0

Compare Source

v1.122.0

Released at 2025-07-18

Update Note 1: vmsingle and vmstorage in VictoriaMetrics cluster: The -snapshotsMaxAge flag default have been changed to 3d. This enables automatic deletion of snapshots older than 3 days. If you want to keep the previous behavior (never automatically deleting snapshots), please set -snapshotsMaxAge=0. See #​9344 for the details.

Update Note 2: all the VictoriaMetrics Enterprise components: -eula command-line flag is skipped when validating VictoriaMetrics enterprise license. Instead, the -license or -licenseFile command-line flags must be used to provide a valid license key. See these docs for configuration examples.

• SECURITY: upgrade Go builder from Go1.24.4 to Go1.24.5. See the list of issues addressed in Go1.24.5.

• SECURITY: upgrade base docker image (Alpine) from 3.22.0 to 3.22.1. See Alpine 3.22.1 release notes.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: enable automatic deletion of snapshots older than 3 days by default. This helps to avoid wasting disk space due to snapshots not being removed in case of backup failure. See #​9344.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: enable metric names stats tracker by default.

• FEATURE: MetricsQL: revert change on respecting staleness markers when calculating rate and increase functions described in #​8891-comment. This change was introduced in v1.120.0.

• FEATURE: vmselect in VictoriaMetrics cluster: Increase idle timeout from 30s to 2m and ensure at least two connections remain open between vmselect and vmstorage. The change smooths out connection usage during traffic spikes. See #​9464.

• FEATURE: dashboards/alert-statistics: add a new dashboard for displaying historical changes in state of alerts generated by vmalert. The new dashboard should help to identify flappy and noisy alerts, or alerts that never fired before.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: reduce CPU usage caused by garbage collection during indexDB lookups on cache misses. See #​9324 for details.

• BUGFIX: all VictoriaMetrics components: properly set process_cpu_cores_available metric to respect both logical CPU cores and cgroup CPU quota. Previously, it could report incorrect values in containerized environments with no cgroup limits. See #​9449. Thanks to the @​fxrlv.

• BUGFIX: vmagent and vmsingle: properly release memory for /target_response API requests. Previously it could used after return and it may lead to data races and potential data loss during scraping. See #​9466. Thanks to the @​fxrlv.

• BUGFIX: vmagent: Prevent panic when re-packing a corrupted block during protocol downgrade. See #​9417.

• BUGFIX: vmselect in VictoriaMetrics cluster: fix handling of /api/v1/notifiers and /api/v1/metadata endpoints; previously, requests to these endpoints returned 400 Bad Request. vmsingle was not affected.

v1.121.0

Compare Source

v1.121.0

Released at 2025-07-04

Update Note 1: The -retryMaxTime flag has been deprecated. Please use -retryMaxInterval flag instead. For more details, see #​9169.

• FEATURE: all the VictoriaMetrics Enterprise components: improve error message when an empty license is provided via the -license or -licenseFile command-line flags. See #​9337 for the details.

• FEATURE: vmagent: add concurrency option to kafka remoteWrite producer. See 9249 issue for details.

• FEATURE: vmagent and Single-node VictoriaMetrics: support label_limit scrape parameter. It sets labels limit for each scraped sample from a target. If limit is exceeded, the scrape will be rejected and target marked with corresponding scrape error. The label_limit can be defined on global level, per job or during relabeling via __label_limit__ label. See scrape configs doc for details. See #​7660 and #​3233 issues.

• FEATURE: vmagent: introduce a new flag -retryMaxInterval as a replacement for the deprecated -retryMaxTime flag. The new flag more accurately reflects the behavior it controls. See #​9169 for more details. Thanks to the @​leiwingqueen.

• FEATURE: vmalert: add /api/v1/notifiers API endpoint for returning list of configured or discovered notifiers.

• FEATURE: vmalert: add datasource_type query argument for /api/v1/rules and /api/v1/alerts endpoints to filter response by rule's datasource type. See #​8537.

• FEATURE: vmalert: respect group order defined in the rule file during replay mode to allow chained group if needed. See #​9334.

• FEATURE: vmalert: add -replay.ruleEvaluationConcurrency to allow concurrent /query_range requests for a single rule. Increasing this value when replaying for a long time and a single request range is limited by -replay.maxDatapointsPerQuery. See this #​7387 for details. Thanks to the @​BenNF.

• FEATURE: vmbackupmanager: improve error messages when vmbackupmanager fails to create snapshot. See #​9340 for the details.

• FEATURE: vmbackupmanager: support client-side TLS configuration for creating and deleting snapshots via -snapshot.tls* cmd-line flags.

• FEATURE: vmsingle and vmagent: remove duplicate kubernetes targets from service-discovery-debug page. See 8626 issue for details.

• FEATURE: vmsingle and vmselect in VictoriaMetrics cluster: add command-line flag -search.logSlowQueryStatsHeaders for query execution stats. The new flag allows specifying the list of headers to log together with slow queries if user's request contains them. This flag is available only in VictoriaMetrics enterprise. See Query Stats for details.

• FEATURE: vmsingle and vmselect: add ability to proxy /api/v1/notifiers to vmalert when -vmalert.proxyURL is set. See 9267 PR for details.

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: add vm_cache_eviction_bytes_total counter metrics to reflect cache evictions due to expiration, misses and cache size. See 9293 PR for details. Thanks to the @​BenNF

• FEATURE: vmsingle and vmstorage in VictoriaMetrics cluster: enhance MustReadAt panic message to include filename for easier debugging of out-of-range reads. See #​9106.

• BUGFIX: dashboards/vmagent and dashboards/vmalert: fix ad-hoc filters auto-complete and filtering on panels that use MetricsQL specific expressions. See #​8657.

• BUGFIX: MetricsQL: fix negative increase result when -search.maxLookback or -search.maxStalenessInterval are set and data contains gap. See #​8935 (comment).

• BUGFIX: stream aggregation: properly clean quantiles output state during flush. See #​9350.

• BUGFIX: vmalert: fix exposition of duplicated metrics for dynamically discovered notifiers via Consul and DNS. See #​9260.

• BUGFIX: vmalert: remove inline styles from UI to align with --http.header.csp=default-src 'self' setting. See #​9236.

• BUGFIX: vmalert: fix alerts state restoration for alerting rules that are using templating in the labels. See #​9305.

• BUGFIX: vmbackup, vmbackupmanager: automatically retry requests failing with Expired Token errors. This helps to avoid failed backups when using EKS Pod Identity for authentication. See #​9280.

• BUGFIX: vmsingle and vmstorage in VictoriaMetrics cluster: properly return results for search requests with .+|^$ regex filter expression. See 9290 issue for details.

• BUGFIX: vmui: fix errors in console about loading of manifest.json when accessing UI through vmauth with Basic Auth enabled.

• BUGFIX: vmui: disabled opening of autocomplete popup on initial page load.

v1.120.0

Compare Source

v1.120.0

Released at 2025-06-20

• SECURITY: upgrade Go builder from Go1.24.3 to Go1.24.4. See the list of issues addressed in Go1.24.4.

• SECURITY: upgrade base docker image (Alpine) from 3.21.3 to 3.22.0. See Alpine 3.22.0 release notes.

• FEATURE: all the VictoriaMetrics components: add -http.disableKeepAlive to disable HTTP keep-alives for incoming connections. The flag could improve load balancing among replicas behind HTTP load balancers. See #​9125 and #​2395 for details.

• FEATURE: dashboards/cluster: add panel Partitions scheduled for re-processing to Troubleshooting row. It shows the amount of data scheduled for downsampling or retention filters. The new panel should help to correlate resource usage with background re-processing of partitions.

• FEATURE: MetricsQL: support rate_prometheus function, an equivalent to increase_prometheus(series_selector[d]) / d. See #​8901 and #​8891 for details.

• FEATURE: MetricsQL: respect staleness markers when calculating rate and increase functions. The new behavior will interrupt rate/increase calculation if last sample on the selected time window is a staleness marker, making the series to disappear immediately instead of slowly fading away. See more details in #​8891-comment.

• FEATURE: vmalert: do not break vmalert process under replay mode when rule uses query template, but only logging a warning.

• FEATURE: [vmalert](https://docs.victori