Release history
VictoriaMetrics releases
VictoriaMetrics: fast, cost-effective monitoring solution and time series database
All releases
39 shown
- Upgrade Go builder to version 1.26.3, addressing security issues listed in the Go 1.26.3 changelog.
Full changelog
v1.136.9
Released at 2026-05-08
v1.136.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.136.x line will be supported for at least 12 months since v1.136.0 release
-
SECURITY: upgrade Go builder from Go1.26.2 to Go1.26.3. See the list of issues addressed in Go1.26.3.
-
BUGFIX: vmagent and vmsingle: properly obtain
__meta_hetzner_hcloud_locationand__meta_hetzner_hcloud_location_network_zonelabels for hetzner_sd_configs. Hetzner changed discovery API response and returnslocationinformation from different field. See #10909. Thanks to @juliusrickert for contribution. -
BUGFIX: vmauth: now correctly respects disabling retries via
-maxRequestBodySizeToRetry=0. Previously, disabling retries required setting both-maxRequestBodySizeToRetryand-requestBufferSizeto0. See #10857. Thanks to @andriibeee for the contribution. -
BUGFIX: vmbackupmanager: explicitly set the MD5 checksum for
DeleteObjectsrequests. Starting with aws-sdk-go-v2/service/s3 v1.73.0, the SDK switched the checksum algorithm from MD5 to CRC32, which can break compatibility with some third-party S3 implementations, including Dell ECS. See #10907. -
BUGFIX: vmauth: attempt to route requests to the first backend when all backends are marked as broken. Previously,
vmauthreturned an error immediately without attempting any backend. This change may improve success rate in rare edge cases. See #10837. -
BUGFIX: vmsingle: and
vmselectin VictoriaMetrics cluster: give priority toextra_labelandextra_filters[]params defined in URL query string over those defined in request body form, to properly respect constraints imposed byvmauth. See #10908. -
BUGFIX: vmsingle and
vmselectin VictoriaMetrics cluster: set CORS headers on/api/v1/export,/api/v1/export/csvand/api/v1/export/native. See #10899. Thanks to @andriibeee for the contribution.
- Go builder upgraded from Go1.26.2 to Go1.26.3 addressing security issues
Full changelog
v1.122.22
Released at 2026-05-08
v1.122.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.122.x line will be supported for at least 12 months since v1.122.0 release
-
SECURITY: upgrade Go builder from Go1.26.2 to Go1.26.3. See the list of issues addressed in Go1.26.3.
-
BUGFIX: vmagent and vmsingle: properly obtain
__meta_hetzner_hcloud_locationand__meta_hetzner_hcloud_location_network_zonelabels for hetzner_sd_configs. Hetzner changed discovery API response and returnslocationinformation from different field. See #10909. Thanks to @juliusrickert for contribution. -
BUGFIX: vmsingle: and
vmselectin VictoriaMetrics cluster: give priority toextra_labelandextra_filters[]params defined in URL query string over those defined in request body form, to properly respect constraints imposed byvmauth. See #10908. -
BUGFIX: vmsingle and
vmselectin VictoriaMetrics cluster: set CORS headers on/api/v1/export,/api/v1/export/csvand/api/v1/export/native. See #10899. Thanks to @andriibeee for the contribution.
- SECURITY: upgrade Go builder from Go1.26.2 to Go1.26.3 (addresses issues listed in https://github.com/golang/go/issues?q=milestone%3AGo1.26.3%20label%3ACherryPickApproved)
- Suppress TCP health check errors when `-tls` flag is set for all components
- Add `__meta_hetzner_robot_datacenter` label in vmagent and vmsingle hetzner_sd_configs
- Introduce `-rule.stripFilePath` option in vmalert
Full changelog
v1.143.0
Released at 2026-05-08
-
SECURITY: upgrade Go builder from Go1.26.2 to Go1.26.3. See the list of issues addressed in Go1.26.3.
-
FEATURE: all VictoriaMetrics components: suppress TCP health check errors when
-tlsflag is set. See #10538. -
FEATURE: vmagent and vmsingle: add
__meta_hetzner_robot_datacenterlabel forrobotrole in hetzner_sd_configs. See #10909. Thanks to @juliusrickert for contribution. -
FEATURE: vmalert: add
-rule.stripFilePathto support stripping rule file paths in logs and all API responses, including /metrics. See #5625. -
FEATURE: vmalert: add
formatTimetemplate function for formatting a Unix timestamp using the provided layout. For example,{{ now | formatTime "2006-01-02T15:04:05Z07:00" }}returns the current time in RFC3339 format. See issue #10624. Thanks to @andriibeee for the contribution. -
FEATURE: vmagent, vmsingle,
vminsertin VictoriaMetrics cluster: add support for Prometheus native histogram during ingestion. See #10743. -
FEATURE: vmagent, vmsingle,
vmselectin VictoriaMetrics cluster andvmstoragein VictoriaMetrics cluster: introduce thevm_fs_infometric. It exposes the filesystem type (e.g., ext4, xfs, nfs) used for-*Pathrelated flags. See #10482. -
FEATURE: dashboards/vmagent: add
Kafka (Enterprise)row with panels for monitoring traffic (bytes), messages in/out, producer and consumer errors. See #10728. -
BUGFIX: vmagent and vmsingle: properly obtain
__meta_hetzner_hcloud_locationand__meta_hetzner_hcloud_location_network_zonelabels for hetzner_sd_configs. Hetzner changed discovery API response and returnslocationinformation from different field. See #10909. Thanks to @juliusrickert for contribution. -
BUGFIX:
vminsertin VictoriaMetrics cluster: properly discover addresses of storage nodes with enterprise automatic-vmstorage-discovery. Bug was introduced at v1.141.0. -
BUGFIX: vmauth: now correctly respects disabling retries via
-maxRequestBodySizeToRetry=0. Previously, disabling retries required setting both-maxRequestBodySizeToRetryand-requestBufferSizeto0. See #10857. Thanks to @andriibeee for the contribution. -
BUGFIX: vmbackupmanager: explicitly set the MD5 checksum for
DeleteObjectsrequests. Starting with aws-sdk-go-v2/service/s3 v1.73.0, the SDK switched the checksum algorithm from MD5 to CRC32, which can break compatibility with some third-party S3 implementations, including Dell ECS. See #10907. -
BUGFIX: vmauth: attempt to route requests to the first backend when all backends are marked as broken. Previously,
vmauthreturned an error immediately without attempting any backend. This change may improve success rate in rare edge cases. See #10837. -
BUGFIX: vmsingle: and
vmselectin VictoriaMetrics cluster: give priority toextra_labelandextra_filters[]params defined in URL query string over those defined in request body form, to properly respect constraints imposed byvmauth. See #10908. -
BUGFIX: vmsingle and
vmselectin VictoriaMetrics cluster: set CORS headers on/api/v1/export,/api/v1/export/csvand/api/v1/export/native. See #10899. Thanks to @andriibeee for the contribution.
Fixed vminsert node discovery in VictoriaMetrics cluster with automatic-vmstorage-discovery.
Full changelog
v1.136.8
Released at 2026-04-30
v1.136.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.136.x line will be supported for at least 12 months since v1.136.0 release
- BUGFIX:
vminsertin VictoriaMetrics cluster: properly discover addresses of storage nodes with enterprise automatic-vmstorage-discovery. Bug was introduced at v1.141.0.
Fixes OpenTelemetry Unit suffix incorrectly applying across metrics.
Full changelog
v1.142.0
Released at 2026-04-28
- BUGFIX: vmagent, vmsingle and
vminsertin VictoriaMetrics cluster: properly add OpenTelemetryUnitsuffix to the metric name and metadataHelpwhen-opentelemetry.usePrometheusNamingflag is enabled. Previously, ifhttp_requestshasUnit: secondsand the next metriccpu_usagehas noUnit,cpu_usagewould be ingested ascpu_usage_seconds. See #10889.
Fixes OpenTelemetry Unit suffix incorrectly applying across metrics.
Full changelog
v1.136.7
Released at 2026-04-28
v1.136.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.136.x line will be supported for at least 12 months since v1.136.0 release
- BUGFIX: vmagent, vmsingle and
vminsertin VictoriaMetrics cluster: properly add OpenTelemetryUnitsuffix to the metric name and metadataHelpwhen-opentelemetry.usePrometheusNamingflag is enabled. Previously, ifhttp_requestshasUnit: secondsand the next metriccpu_usagehas noUnit,cpu_usagewould be ingested ascpu_usage_seconds. See #10889.
- Alpine base image upgraded from 3.23.3 to 3.23.4
- Systemd slices resource limit detection
- Future timestamp support with -futureRetention flag
- Stream aggregation counter reset metrics
Full changelog
v1.141.0
Released at 2026-04-24
-
SECURITY: upgrade base docker image (Alpine) from 3.23.3 to 3.23.4. See Alpine 3.23.4 release notes.
-
FEATURE: all VictoriaMetrics components: add support for reading cpu/memory limits configured via systemd slices. Previously, only limits set directly on the process's own cgroup were detected. See #10635. Thanks to @andriibeee for the contribution.
-
FEATURE: vmctl: improve error handling at opentsdb migration. See #10797
-
FEATURE: vmui: now
Run querylink on the Alerting Rules page correctly propagates the alert’s interval and evaluation time. See #10366. -
FEATURE: alerts: add new
MetricNameStatsCacheUtilizationIsTooHighalerting rule to track overutilization of Metric names usage stats tracker (used in Cardinality Explorer). See #10840. -
FEATURE: stream aggregation: add
vm_streamaggr_counter_resets_totalmetric fortotal*,increase*andrate*outputs that is useful for aggregation behaviour tracking. These metrics help to identify issues described in Troubleshooting: counter resets. See #10807. -
FEATURE: vmsingle and
vmstorageandvmselectin VictoriaMetrics cluster: add the support of ingestion and retrieval of samples with timestamps in the future. The new-futureRetentionflag controls how far in the future the timestamps are allowed to be. See #827 and #10718. -
FEATURE: dashboards/vmagent, dashboards/cluster: add metadata row ingestion rate queries to related panels. See #10868.
-
BUGFIX:
vminsertin VictoriaMetrics cluster: fix increased memory usage after upgrade to v1.140.0 by properly accounting for internal buffer count when calculating per-storage buffer size. See #10725. Thanks to @fxrlv for the contribution. -
BUGFIX: all VictoriaMetrics components: properly parse IPv6 source address when accepting connections with proxy protocol v2 enabled. See #10839. Thanks to @andriibeee for the contribution.
-
BUGFIX: vmagent: properly attach tenant from
__tenant_id__label to the scraped metadata. See #10828. -
BUGFIX: vmagent: keep tenant ingested into vmagent via prometheus remotewrite endpoint. See #10828.
-
BUGFIX: vmagent and vmsingle:
-maxScrapeSizeis now correctly applied when reading response bodies, including non-OK scrape error responses. See #10804. -
BUGFIX: vmagent: fix
ec2_sd_configsreturning 401AuthFailurefrom AWS when credentials are obtained via IRSA, instance role orAWS_CONTAINER_CREDENTIALS_*env vars. The regression was introduced in v1.140.0. See #10815. Thanks to @andriibeee for the contribution. -
BUGFIX: vmauth: fix leak of backend TCP connections, file descriptors and goroutines when the client cancels the request after the backend response has been received. See #10833. Thanks to @andriibeee for the contribution.
-
BUGFIX: vmauth: fix a rare panic during config reload when a backend is marked as broken. See #10806.
-
BUGFIX:
vmselectin VictoriaMetrics cluster: stop logging warnings about failed handshakes when theclusternativeport receives TCP healthchecks from load balancers. See #10786. Thanks to @andriibeee for the contribution. -
BUGFIX: vmrestore: fix an issue where vmrestore could hang indefinitely when interrupted during backup download. See #10794.
-
BUGFIX: vmsingle: properly execute graceful shutdown for vmsingle if
-maxIngestionRateis configured. See #10795. -
BUGFIX: vmui: fix time display on Alerting Rules page to use selected timezone. See #10827.
-
BUGFIX: vmui: use contrasting colors when displaying time series to improve visibility on light and dark themes. See #10869.
-
BUGFIX: vmalert: delete labels from rule results if they are specified with an empty string value in rule or group labels. See #10766.
-
BUGFIX: vmsingle and
vmselectin VictoriaMetrics cluster: fix incorrect evaluation of binary operations caused by an ordering bug (e.g.10 - (3 + 3 + 4)being evaluated as10 - 3 + 3 + 4). The issue was introduced in v1.140.0, v1.136.4, and v1.122.19. See #10856.
- Go builder upgraded from 1.25.9 to 1.26.2
- Alpine base image upgraded from 3.23.3 to 3.23.4
Full changelog
v1.122.21
Released at 2026-04-24
v1.122.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.122.x line will be supported for at least 12 months since v1.122.0 release
-
SECURITY: upgrade Go builder from Go1.25.9 to Go1.26.2. See the list of issues addressed in Go1.26.2.
-
SECURITY: upgrade base docker image (Alpine) from 3.23.3 to 3.23.4. See Alpine 3.23.4 release notes.
-
BUGFIX: all VictoriaMetrics components: properly parse IPv6 source address when accepting connections with proxy protocol v2 enabled. See #10839. Thanks to @andriibeee for the contribution.
-
BUGFIX: vmagent and vmsingle:
-maxScrapeSizeis now correctly applied when reading response bodies, including non-OK scrape error responses. See #10804. -
BUGFIX:
vminsertin VictoriaMetrics cluster: properly allocate vminsert buffer size per vmstorage node based on available CPU, memory and storage node count to reduce OOM risk. See #10725. -
BUGFIX: vmrestore: fix an issue where vmrestore could hang indefinitely when interrupted during backup download. See #10794.
-
BUGFIX: vmsingle: properly execute graceful shutdown for vmsingle if
-maxIngestionRateis configured. See #10795. -
BUGFIX: vmalert: delete labels from rule results if they are specified with an empty string value in rule or group labels. See #10766.
-
BUGFIX: vmui: use contrasting colors when displaying time series to improve visibility on light and dark themes. See #10869.
- Alpine base image upgraded from 3.23.3 to 3.23.4
Full changelog
v1.136.6
Released at 2026-04-24
v1.136.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.136.x line will be supported for at least 12 months since v1.136.0 release
-
SECURITY: upgrade base docker image (Alpine) from 3.23.3 to 3.23.4. See Alpine 3.23.4 release notes.
-
BUGFIX: all VictoriaMetrics components: properly parse IPv6 source address when accepting connections with proxy protocol v2 enabled. See #10839. Thanks to @andriibeee for the contribution.
-
BUGFIX: vmagent: properly attach tenant from
__tenant_id__label to the scraped metadata. See #10828. -
BUGFIX: vmagent: keep tenant ingested into vmagent via prometheus remotewrite endpoint. See #10828.
-
BUGFIX: vmagent and vmsingle:
-maxScrapeSizeis now correctly applied when reading response bodies, including non-OK scrape error responses. See #10804. -
BUGFIX: vmagent: fix
ec2_sd_configsreturning 401AuthFailurefrom AWS when credentials are obtained via IRSA, instance role orAWS_CONTAINER_CREDENTIALS_*env vars. The regression was introduced in v1.140.0. See #10815. Thanks to @andriibeee for the contribution. -
BUGFIX: vmauth: fix leak of backend TCP connections, file descriptors and goroutines when the client cancels the request after the backend response has been received. See #10833. Thanks to @andriibeee for the contribution.
-
BUGFIX: vmauth: fix a rare panic during config reload when a backend is marked as broken. See #10806.
-
BUGFIX:
vmselectin VictoriaMetrics cluster: stop logging warnings about failed handshakes when theclusternativeport receives TCP healthchecks from load balancers. See #10786. Thanks to @andriibeee for the contribution. -
BUGFIX: vmrestore: fix an issue where vmrestore could hang indefinitely when interrupted during backup download. See #10794.
-
BUGFIX: vmsingle: properly execute graceful shutdown for vmsingle if
-maxIngestionRateis configured. See #10795. -
BUGFIX: vmui: fix time display on Alerting Rules page to use selected timezone. See #10827.
-
BUGFIX: vmui: use contrasting colors when displaying time series to improve visibility on light and dark themes. See #10869.
-
BUGFIX: vmalert: delete labels from rule results if they are specified with an empty string value in rule or group labels. See #10766.
Critical bugfix: corrects binary operation evaluation ordering in vmsingle and vmselect.
Full changelog
v1.122.20
Released at 2026-04-23
v1.122.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.122.x line will be supported for at least 12 months since v1.122.0 release
- BUGFIX: vmsingle and
vmselectin VictoriaMetrics cluster: fix incorrect evaluation of binary operations caused by an ordering bug (e.g.10 - (3 + 3 + 4)being evaluated as10 - 3 + 3 + 4). The issue was introduced in v1.140.0, v1.136.4, and v1.122.19. See #10856.
Critical bugfix: corrects binary operation evaluation ordering in vmsingle and vmselect.
Full changelog
v1.136.5
Released at 2026-04-23
v1.136.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.136.x line will be supported for at least 12 months since v1.136.0 release
- BUGFIX: vmsingle and
vmselectin VictoriaMetrics cluster: fix incorrect evaluation of binary operations caused by an ordering bug (e.g.10 - (3 + 3 + 4)being evaluated as10 - 3 + 3 + 4). The issue was introduced in v1.140.0, v1.136.4, and v1.122.19. See #10856.
- CSV export `/api/v1/export/csv` now includes header row as first line; existing CSV-processing scripts may need to skip this header
- Go builder upgraded from Go1.26.1 to Go1.26.2
- vmctl thanos mode for migrating Thanos data with raw and downsampled blocks
- vmagent per-URL `-remoteWrite.disableMetadata` flag
- EC2 service discovery with AWS profile support including source_profile and role_arn resolution
- Go builder upgraded from Go1.26.1 to Go1.26.2
- Go builder upgraded from Go1.25.8 to Go1.25.9
Fixed RPC connection reuse issues after failed queries, improved Grafana Prometheus alert proxying, fixed UI autocomplete dropdown closing.
Fixed RPC connection pooling for query resilience, corrected UI autocomplete dropdown closure, improved Grafana Prometheus alert proxying, fixed OpenMetrics type logging, and prevented NFS directory deletion panics.
- seriesCountByMetricName table in Cardinality Explorer
- Unused metrics Grafana dashboard
- Search and pagination in vmalert API
- Upgraded Go from Go1.24.13 to Go1.25.8
- histogram_fraction function in MetricsQL for bucket analysis
- Go builder upgraded from 1.26.0 to 1.26.1
- histogram_fraction MetricsQL function
- Disabled Graphite tag registration endpoints
Full changelog
v1.136.2
Released at 2026-03-13
v1.136.x is a line of LTS releases. It contains important up-to-date bugfixes for VictoriaMetrics enterprise.
All these fixes are also included in the latest community release.
The v1.136.x line will be supported for at least 12 months since v1.136.0 release
-
SECURITY: upgrade Go builder from Go1.26.0 to Go1.26.1. See the list of issues addressed in Go1.26.1.
-
FEATURE: vmsingle and
vmselectin VictoriaMetrics cluster: Disable/graphite/tags/tagSeriesand/graphite/tags/tagMultiSeriesfor Graphite tag registration since it is unlikely it is used in context of VictoriaMetrics. See 10544. -
FEATURE: MetricsQL: add histogram_fraction function to calculate the fraction of buckets falling between lowerLe and upperLe. See #5346.
-
BUGFIX: all VictoriaMetrics components: replace
histogramwithuntypedmetric metadata type for VictoriaMetrics histograms when-metrics.exposeMetadatais set. See #82. -
BUGFIX: vmauth: properly route requests to
default_url. Previously,request_pathquery arg could be set incorrectly during concurrent requests. See #10626. -
BUGFIX: vmbackup, vmbackupmanager: do not enable ACL when uploading backups to S3-compatible endpoints by default. ACL is not always supported by S3-compatible endpoints and it is not recommended to use ACLs to limit access to objects. See #10539 for more details.
-
BUGFIX: vmbackupmanager: overwrite s3 object metadata while syncing latest backups with other backup types. See #10639.
-
BUGFIX: vmagent, vmsingle,
vminsertandvmstoragein VictoriaMetrics cluster: properly attachhostlabel to the time series ingested via /datadog/api/beta/sketches API. See #10557. -
BUGFIX: vmui: use
increase_pureinstead ofratefor histogram heatmaps in Explore Metrics to correctly display the first observation in each new bucket. See #10365. Thanks to @ab0utbla-k for the contribution. -
BUGFIX:
vmstoragein VictoriaMetrics cluster: fix inaccuratevm_filestream_write_duration_seconds_totaldue to duplicate counting . After the fix,vm_filestream_write_duration_seconds_totalwill track the duration spent on calling thewrite(2)system call properly. See #10564.
- Upgraded Go from Go1.26.0 to Go1.26.1
- OIDC Discovery for JWT authentication
- OAuth2 custom headers support
- Per-user access logging
Fixed gzip compression preference over zstd, mTLS certificate expiration checking, regex nesting panic attacks, multitenant label filtering, and instant query rollup cache regression.
Fixed tenant result filtering in multitenant query requests for proper tenant isolation.
Fixed regex nesting panics from large repetition ranges, improved multitenant tenant searching, corrected extra_filters application in label queries, and fixed UI label name escaping.
- vmbackup cross-storage backend copying
- JWT token authentication with signature verification
- Dynamic URL rewriting using JWT claim fields
- Upgraded Alpine from 3.23.2 to 3.23.3
- Upgraded Go from Go1.24.12 to Go1.24.13
- Upgraded Alpine from 3.23.2 to 3.23.3
- Upgraded Go from Go1.24.12 to Go1.24.13
- Upgraded Go from Go1.25.6 to Go1.26.0
- Upgraded Alpine from 3.23.2 to 3.23.3
- Clickable source code links in dashboards
- Queries with most memory to execute section
- Context-aware label value autocomplete
Fixed MetricsQL changes() function handling gaps, alert restore for slow group evaluation, vmauth backend health check spam during reloads, tenant filtering in multitenant queries, and cardinality explorer percentage calculations.
Fixed vmauth backend health check spam during config reloads, corrected tenant result filtering, improved cardinality explorer percentage calculations, fixed decompressed scrape size checking, and resolved alert restore issues.
- Request body buffering in vmauth
- Per-URL queue configuration in vmagent
- New alerting rules for persistent queue capacity
- Upgraded Go from Go1.24.11 to Go1.24.12
- CVE-2025-61729 - upgraded Go from Go1.24.11 to Go1.24.12
- Upgraded Go from Go1.25.5 to Go1.25.6
- vm_persistentqueue_free_disk_space_bytes metric
- Month (M) duration unit for retention
- Persistent queue Full ETA dashboard
- CVE-2025-61729 - upgraded Go from Go1.24.10 to Go1.24.11
- Upgraded Alpine from 3.22.2 to 3.23.2
- CVE-2025-61729 - upgraded Go from Go1.24.10 to Go1.24.11
- Upgraded Alpine from 3.22.2 to 3.23.2
- Per-partition index upgrade requires registering all time series and may cause temporary slowdown
- Per-partition index may increase disk usage for retention periods shorter than 1 month
- Certain data patterns may cause high CPU utilization
- Upgraded Alpine from 3.22.2 to 3.23.2
- Global sampleLimit setting
- Per-partition index for improved scale
- Enhanced metrics tracking