profile
viewpoint

johnatannvmd/codeception-mailchecker-module 2

Codeception MailChecker module. Was made for loving yours email acceptance tests.

johnatannvmd/johnatan.ru 0

johnatan.ru sources

johnatannvmd/VictoriaMetrics 0

VictoriaMetrics: fast, cost-effective monitoring solution and time series database

push eventjohnatannvmd/VictoriaMetrics

Aliaksandr Valialkin

commit sha 2e237a6e8c2d89149fcf9d31633abe58dcaef1fd

docs/CHANGELOG.md: document b9cdbcb5046315db96e1e7ca9923d09d0f30dc25

view details

Aliaksandr Valialkin

commit sha 7cbb81c2a6e3e25cc24d761f6ee0ac4e45eae2a0

Makefile: add `TAG=v... make publish-release` rule for building and publishing a release for the given TAG

view details

Aliaksandr Valialkin

commit sha 62edbcdb23366ba3e3eb22c2a975221f4fa79922

docs/MetricsQL.md: clarify documentation for `lifetime` function Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1780

view details

Aliaksandr Valialkin

commit sha d59acb06c1c8759ca651c176a675053ea848bc32

vendor: `make vendor-update`

view details

Aliaksandr Valialkin

commit sha 3a1ac218baa7b465e0abe699f2c05ea7eaebd5c8

app/vmselect/promql: add `duration_over_time(m[d], max_interval)` function This function calculates the actual lifetime of the time series on the given lookbehdind window `d` Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1780

view details

Aliaksandr Valialkin

commit sha f41c02e47563456f3ee31f48acbdd39036e9ac4f

lib/promscrape: improve logging for `scrape_config_files` parse errors Log the actual file path, which led to the parse error. Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1789

view details

Yury Molodov

commit sha 89e0f66bfc7a50ccce91094919803904aaa44d52

vmui: migration MUI Core v4 to v5 (#1795) * migration MUI Core v4 to v5 * app/vmui: `make vmui-update` Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>

view details

Aliaksandr Valialkin

commit sha b1de180886b56345b3455926f4fe966fb624ca1a

docs/CHANGELOG.md: cut v1.69.0

view details

Aliaksandr Valialkin

commit sha a352b59b7ea92c1825677d049bb28f3e752f883e

docs/MetricsQL.md: mention than tlast_over_time() is an alias for timestamp()

view details

Aliaksandr Valialkin

commit sha 57aaf914acc848e4684a88d85f821c1f0af1920b

docs: mention that graphs on the official dashboards contain useful hints

view details

Aliaksandr Valialkin

commit sha 31486f224424b45e7dbe4fe0511fd18ee6ebcd22

app/vmselect/promql: properly return durations smaller than one second from duration_over_time() function Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1780

view details

Aliaksandr Valialkin

commit sha bc357d7132c25dae22b144107a25fdd0e7ebf99a

docs/vmalert.md: improve wording in `Multitenancy` chapter

view details

Aliaksandr Valialkin

commit sha 6346b78fa89d546b1d8db9b3a215c5a396d259fd

dashboards: consistently use regexp filters for template vars (#1799) Template vars may contain regexp when `all` is selected (.*) or when multiple values are selected (foo|bar). So they must be passed to regexp filters.

view details

Aliaksandr Valialkin

commit sha 4fb19fe34bd2b013e3d4659d1b4ea3f89371c919

all: consistently return `application/json` content-type without `charset=utf-8` The `application/json` content-type has utf-8 encoding by default. See https://stackoverflow.com/questions/9254891/what-does-content-type-application-json-charset-utf-8-really-mean Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/897

view details

Aliaksandr Valialkin

commit sha 305507930c6822aabf3ea411c4a3444cb0167984

app/vmauth: initialize reverse proxy only after flag.Parse() is called This should properly take into accoun the `-maxIdleConnsPerBackend` command-line flag value. Previously it was hardcoded to 100. Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1300

view details

Aliaksandr Valialkin

commit sha fd4d665079883bef21914d1aa3e85c7816fb2ef5

docs/Single-server-VictoriaMetrics.md: mention that it is possible to send gzipped data to /api/v1/import/prometheus

view details

Aliaksandr Valialkin

commit sha 82917398b9fe02e6061f653fb9596a246252be9e

docs/CHANGELOG.md: document bugfixes in enteprise versions of vmagent and vmalert

view details

Aliaksandr Valialkin

commit sha 82cf9f26db6df9528e4cd65b2f171a525486d058

app/vmselect/promql: prevent from incorrect calculations for `deriv()` over multiple samples with identical timestamps

view details

Aliaksandr Valialkin

commit sha 471c0ddd88b6a79eb04c008c7c1d588e5c81fda5

docs/Articles.md: add a linkt to OSA Con talk about `how clickhouse inspired us to build victoriametrics`

view details

Aliaksandr Valialkin

commit sha 8cb7074030f18ec75199f57bf76791996f06b89c

deployment/docker: allow using `/` chars in `ROOT_IMAGE` when running `make package-*` This fixes the following command: ROOT_IMAGE=gcr.io/distroless/static make package-victoria-metrics

view details

push time in 3 days

push eventjohnatannvmd/VictoriaMetrics

Aliaksandr Valialkin

commit sha 5edf695bc971c4baf4c8337bd4fe451ea85b743a

docs/CHANGELOG.md: document b9cdbcb5046315db96e1e7ca9923d09d0f30dc25

view details

Aliaksandr Valialkin

commit sha b07010839c7fc1c43fd5d18f520de88f300666a8

Makefile: add `TAG=v... make publish-release` rule for building and publishing a release for the given TAG

view details

Aliaksandr Valialkin

commit sha e44137d46b979e9a54a31b3ea947dbd0f833aa67

docs/MetricsQL.md: clarify documentation for `lifetime` function Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1780

view details

Aliaksandr Valialkin

commit sha 1be4838ca06e28e5da2fba0d730839871312fe42

vendor: `make vendor-update`

view details

Aliaksandr Valialkin

commit sha 3419ac1d36b54ca394926a36b7e72077998ed73c

app/vmselect/promql: add `duration_over_time(m[d], max_interval)` function This function calculates the actual lifetime of the time series on the given lookbehdind window `d` Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1780

view details

Aliaksandr Valialkin

commit sha fd596945e7332f3d0927ad0cbae78a85fa89a2ec

lib/promscrape: improve logging for `scrape_config_files` parse errors Log the actual file path, which led to the parse error. Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1789

view details

Yury Molodov

commit sha b78ab88a1c1d349eabe43ad70996aa426e29eb1f

vmui: migration MUI Core v4 to v5 (#1795) * migration MUI Core v4 to v5 * app/vmui: `make vmui-update` Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>

view details

Aliaksandr Valialkin

commit sha 69e655ba7f9a9b0d096feec638a89e0b4074c18b

docs/CHANGELOG.md: cut v1.69.0

view details

Aliaksandr Valialkin

commit sha f39ee8dc957f61361e451feff09adae59dc4106a

docs/MetricsQL.md: mention than tlast_over_time() is an alias for timestamp()

view details

Aliaksandr Valialkin

commit sha a67518fc6df7337f414c5da1583c8d3a7430d173

docs: mention that graphs on the official dashboards contain useful hints

view details

Aliaksandr Valialkin

commit sha 840ac283ef4eaf1eb9b56e46b2bc016d653ca3da

app/vmselect/promql: properly return durations smaller than one second from duration_over_time() function Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1780

view details

Aliaksandr Valialkin

commit sha 5046efb94b4c3919c79ed84b825e646d639d135b

docs/vmalert.md: improve wording in `Multitenancy` chapter

view details

Aliaksandr Valialkin

commit sha 802f05f73f9efc25549a8a83d4c19ff6d5357d83

dashboards: consistently use regexp filters for template vars (#1798) Template vars may contain regexp when `all` is selected (.*) or when multiple values are selected (foo|bar). So they must be passed to regexp filters.

view details

Aliaksandr Valialkin

commit sha e5ac9d8e57fa8a22bc18185573207d71087a2027

all: consistently return `application/json` content-type without `charset=utf-8` The `application/json` content-type has utf-8 encoding by default. See https://stackoverflow.com/questions/9254891/what-does-content-type-application-json-charset-utf-8-really-mean Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/897

view details

Aliaksandr Valialkin

commit sha e5d4c7f4a7fd9bb9957dddd58a5776bc525e15b2

app/vmauth: initialize reverse proxy only after flag.Parse() is called This should properly take into accoun the `-maxIdleConnsPerBackend` command-line flag value. Previously it was hardcoded to 100. Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1300

view details

Aliaksandr Valialkin

commit sha e2f823fffc0ecc136658b8c496335a3390ce5233

docs/Single-server-VictoriaMetrics.md: mention that it is possible to send gzipped data to /api/v1/import/prometheus

view details

Aliaksandr Valialkin

commit sha 8b6c89423d8feb9bd11c645679d868b919120cc4

docs/CHANGELOG.md: document bugfixes in enteprise versions of vmagent and vmalert

view details

Aliaksandr Valialkin

commit sha 9fa098d8e3fbb4bab19d92210bcb325b0ee1bb04

app/vmselect/promql: prevent from incorrect calculations for `deriv()` over multiple samples with identical timestamps

view details

Aliaksandr Valialkin

commit sha c77ff2d293b65dc0ab8b243e51dedc0812b686a6

docs/Articles.md: add a linkt to OSA Con talk about `how clickhouse inspired us to build victoriametrics`

view details

Denys Holius

commit sha 49ee952e9a79ad66e8a1f1309b7ac9e3a99363d5

Bumped Alpine linux version to the latest (#1811) See this https://alpinelinux.org/posts/Alpine-3.14.3-released.html

view details

push time in 3 days

issue commentVictoriaMetrics/VictoriaMetrics

InfluxDB metrics

@aidastefaniamanole, or you can send so called "measurementless" metrics(if library supports this) like this:

,tag1=value1,tag2=value2 value1=12,value2=40

and VictoriaMetrics will save yours metrics like this:

field1{tag1="value1", tag2="value2"} 12
field2{tag1="value1", tag2="value2"} 40
aidastefaniamanole

comment created time in 3 days

startedgoccmack/gocc

started time in 5 days

startedlapce/lapce

started time in 13 days

startedgo-faster/ch

started time in 22 days

issue commentlunasec-io/lunasec

False positive Log4J Checks - JndiLookup.class

Can confirm this

pathetiq

comment created time in a month

startedtidwall/uhaha

started time in 2 months

issue closedVictoriaMetrics/VictoriaMetrics

resets does not work as expected

Describe the bug I have several service that expose uptime metrics. This is a counter, which increments every second. Merics gathers every minute.

RAW metrics look like this: Screenshot from 2021-11-19 18-35-48 Application health-data-as-seriestocolumns-2021-11-19 15_37_55.csv - there are 6 hosts

A want to know when service is restarted. So I applied resets function with 1m lookbehind and got something like this: Screenshot from 2021-11-19 18-41-19 Which looks not quite right. Same result I got when remove lookbehind window and leave that to VM, according to the article this should work. https://valyala.medium.com/promql-tutorial-for-beginners-9ab455142085

... where [d] may be omitted — in this case it equals to the duration between two subsequent points on the graph (aka step):

To Reproduce Import given csv, and make queries.

Expected behavior Resets should return 1 on counter resets.

Version cluster 1.67.0

Command line

vmselect -search.latencyOffset 0 \
-cacheDataPath /data/victora/vmselect_autogen \
-httpListenAddr :8481 \
-replicationFactor 2 \
-search.maxQueryDuration 30s \
-dedup.minScrapeInterval 10s \
-search.logSlowQueryDuration 5s \
-search.resetCacheAuthKey ASDFQWER \
...storages and selectors...
-loggerDisableTimestamps \
-memory.allowedPercent 20
vmstorage -dedup.minScrapeInterval 10s \
-forceFlushAuthKey ASDFQWER \
-forceMergeAuthKey ASDFQWER \
-http.shutdownDelay 5s \
-httpListenAddr :8482 \
-retentionPeriod 1 \
-snapshotAuthKey ASDFQWER \
-storageDataPath /data/victora/autogen \
-vminsertAddr :8400 \
-vmselectAddr :8401 \
-loggerDisableTimestamps \
-search.maxUniqueTimeseries 10000001 \
-memory.allowedPercent 60

closed time in 2 months

johnatannvmd

issue commentVictoriaMetrics/VictoriaMetrics

resets does not work as expected

To achieve my goal, I need to change lookbehind window to 5m and set min step 1m explicitly.

Screenshot from 2021-11-19 19-52-48_cut

Seems like this is my fault, sorry for inconvenience.

johnatannvmd

comment created time in 2 months

issue openedVictoriaMetrics/VictoriaMetrics

resets does not work as expected

Describe the bug I have several service that expose uptime metrics. This is a counter, which increments every second. Merics gathers every minute.

RAW metrics look like this: Screenshot from 2021-11-19 18-35-48 Application health-data-as-seriestocolumns-2021-11-19 15_37_55.csv - there are 6 hosts

A want to know when service is restarted. So I applied resets function with 1m lookbehind and got something like this: Screenshot from 2021-11-19 18-41-19 Which looks not quite right. Same result I got when remove lookbehind window and leave that to VM, according to the article this should work. https://valyala.medium.com/promql-tutorial-for-beginners-9ab455142085

... where [d] may be omitted — in this case it equals to the duration between two subsequent points on the graph (aka step):

To Reproduce Import given csv, and make queries.

Expected behavior Resets should return 1 on counter resets.

Version cluster 1.67.0

Command line

vmselect -search.latencyOffset 0 \
-cacheDataPath /data/victora/vmselect_autogen \
-httpListenAddr :8481 \
-replicationFactor 2 \
-search.maxQueryDuration 30s \
-dedup.minScrapeInterval 10s \
-search.logSlowQueryDuration 5s \
-search.resetCacheAuthKey ASDFQWER \
...storages and selectors...
-loggerDisableTimestamps \
-memory.allowedPercent 20
vmstorage -dedup.minScrapeInterval 10s \
-forceFlushAuthKey ASDFQWER \
-forceMergeAuthKey ASDFQWER \
-http.shutdownDelay 5s \
-httpListenAddr :8482 \
-retentionPeriod 1 \
-snapshotAuthKey ASDFQWER \
-storageDataPath /data/victora/autogen \
-vminsertAddr :8400 \
-vmselectAddr :8401 \
-loggerDisableTimestamps \
-search.maxUniqueTimeseries 10000001 \
-memory.allowedPercent 60

created time in 2 months

startedhectorm/docker-qemu-win2000

started time in 2 months

issue commentVictoriaMetrics/VictoriaMetrics

vmselect return connection pool errors on storage down

Looks like idea of separate timeouts for queries and storages, plus health checks between vmselect and vmstorage already mentioned here: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711#issuecomment-768305643

Also this would be great if we have storages' health metrics on the vmselect/vminsert side

johnatannvmd

comment created time in 2 months

fork johnatannvmd/VictoriaMetrics

VictoriaMetrics: fast, cost-effective monitoring solution and time series database

https://victoriametrics.com/

fork in 3 months

issue commentVictoriaMetrics/VictoriaMetrics

vmselect return connection pool errors on storage down

@valyala Thanks for your response!

VictoriaMetrics doesn't allow returning partial responses for data export queries if some of the vmstorage nodes aren't available Is there any query parameters like deny_partial_response but to accept of partial response for /api/v1/query request?

could you give more context for your use case in order to understand how the exported data is used later and whether it is OK if the exported data may miss a part of samples and series? I do not export data in any cases, "Grafana response" is the error that Grafana shows me when I try to open some dashboards.

We built setup like this to achieve cross location replication(for redundancy) and HA.

download (2) We heavily using tenants to separate metrics across internal teams, so replacement kafka with vmagent is not possible. Tennant IDs is the same across all locations.

Recently we have destruction testing where one of the location was air gapped for one hour. And we got errors that I mention above, during the test. Some of the queries works, some of them not.

The error when exporting data for query error is emitted when the query to /api/v1/query endpoint contains the duration in square brackets at the end I additionally checked the reported errors and there are all kind of queries, with and without square brackets.

Most of all I worry about this part of the message:

cannot obtain connection from a pool: cannot dial 10.120.137.211:8401: dial tcp4 10.120.137.211:8401: i/o timeout

Setup promxy for every tennant will be painful.

johnatannvmd

comment created time in 3 months

issue commentVictoriaMetrics/VictoriaMetrics

vmselect return connection pool errors on storage down

Also this would be great if there was two separate options for timeouts. One for the query timeout, another for the storage timeout. First one will work when connection is good, but query is slow. Second one will work when connection between vmselect and vmstorage is unresponsive.

johnatannvmd

comment created time in 3 months

issue openedVictoriaMetrics/VictoriaMetrics

vmselect return connection pool errors on storage down

Describe the bug We use vmselect for cross cluster search as recommended solution. When one of the storage nodes fails, vmselect return cannot obtain connection from a pool instead of partial response.

image

To Reproduce vmselect with several storage nodes(6 for example), turn off 3 of them.

Expected behavior vmselect should remove down nodes from the pool, and return it back, when storage nodes became available again.

Logs Grafana response

"error when exporting data for query=\"...<REDACTED>..\": error occured during search: cannot fetch query results from vmstorage nodes: cannot perform search on vmstorage 10.120.137.211:8401: cannot obtain connection from a pool: cannot dial 10.120.137.211:8401: dial tcp4 10.120.137.211:8401: i/o timeout"

vmselect logs:

Oct 30 06:43:41 sjc01-c01-vic04 vmselect[6994]: warn        victoriametrics/app/vmselect/main.go:523        error in "/select/18/prometheus/api/v1/query?query=...<REDACTED>...": error when exporting data for query="...<REDACTED>..." on the time range (start=1635575980001, end=1635576160000): cannot fetch data for "AccountID=18, ProjectID=0, MinTimestamp=2021-10-30 06:39:40.001 +0000 UTC, MaxTimestamp=2021-10-30 06:42:40 +0000 UTC, TagFilters=[{...<REDACTED>..., IsNegative: false, IsRegexp: true}\n]": error occured during search: cannot fetch query results from vmstorage nodes: cannot perform search on vmstorage 10.120.137.210:8401: cannot obtain connection from a pool: cannot dial 10.120.137.210:8401: dial tcp4 10.120.137.210:8401: i/o timeout

Version 1.63.0

Used command-line flags -search.latencyOffset 0
-cacheDataPath /data/victoria/vmselect_autogen
-httpListenAddr :8481
-replicationFactor 2
-search.maxQueryDuration 30s
-dedup.minScrapeInterval 10s
-search.logSlowQueryDuration 5s
-search.resetCacheAuthKey ASDFQWER
-selectNode 6 nodes
-storageNode 12 nodes
-loggerDisableTimestamps
-memory.allowedPercent 20

created time in 3 months

more