apitech/loki - loki - Gitea: Git with a cup of tea

Commit Graph

Author	SHA1	Message	Date
Danny Kopping	8fa4134ef9	Update dskit (#10091 )	3 years ago
Dylan Guedes	d1be9d54ed	Change "tenant" to "org_id" (#10017 ) What this PR does / why we need it: Change our writefailures module to use "org_id" instead of "tenant" to be more consistent with the rest of our system.	3 years ago
Vladyslav Diachenko	be1a9c4ec5	Revert "updated to the latest version of dskit" (#9953 ) Reverts grafana/loki#9920	3 years ago
Dylan Guedes	9ed7fa9df4	WriteFailures: Add metrics to increase monitoring (#9904 ) What this PR does / why we need it: Add metrics for our write failures manager. This way we can easily monitor how much logging we're doing and how much we're discarding. Which issue(s) this PR fixes: N/A	3 years ago
Vladyslav Diachenko	3243676f4c	updated to the latest version of dskit (#9920 ) What this PR does / why we need it: Updated Loki with the latest version of dskit Checklist - [ ] Reviewed the [`CONTRIBUTING.md`](https://github.com/grafana/loki/blob/main/CONTRIBUTING.md) guide (required) - [ ] Documentation added - [ ] Tests updated - [ ] `CHANGELOG.md` updated - [ ] If the change is worth mentioning in the release notes, add `add-to-release-notes` label - [ ] Changes that require user attention or interaction to upgrade are documented in `docs/sources/upgrading/_index.md` - [ ] For Helm chart changes bump the Helm chart version in `production/helm/loki/Chart.yaml` and update `production/helm/loki/CHANGELOG.md` and `production/helm/loki/README.md`. [Example PR](`d10549e3ec`) --------- Signed-off-by: Vladyslav Diachenko <vlad.diachenko@grafana.com>	3 years ago
Hamish Forbes	d0baa40df3	Only modify identical timestamps when increment-duplicate-timestamps enabled (#9437 ) When `-validation.increment-duplicate-timestamps` is enabled out of order log entries within the same push request can have their timestamps incorrectly mutated. The code is currently just looking for 2 consecutive entries where the timestamp is not increasing. It should only affect entries where the timestamp is a duplicate Test push request payload: ```json { "streams": [ { "stream": { "increment_duplicate_timestamp": "fixed\|true\|false" }, "values": [ [ "1683665111628000000", "1a: 1683665111628000000" ], [ "1683665111628000000", "1b: 1683665111628000000" ], [ "1683665111628000003", "3: 1683665111628000003" ], [ "1683665111628000002", "2: 1683665111628000002" ] ] } ] } ```` As only the first 2 of these timestamps are the same I would expect all but `1b` to have the timestamps recorded as-is and the logs returned by a query in the order: 1a,1b,2,3. Actual Results: ``` > logcli query --forward --limit 100 -z UTC -q --no-labels --since 60m -o jsonl '{increment_duplicate_timestamp="fixed"}' \| jq { "line": "1a: 1683665111628000000", "timestamp": "2023-05-09T20:45:11.628Z" } { "line": "1b: 1683665111628000000", "timestamp": "2023-05-09T20:45:11.628000001Z" } { "line": "2: 1683665111628000002", "timestamp": "2023-05-09T20:45:11.628000002Z" } { "line": "3: 1683665111628000003", "timestamp": "2023-05-09T20:45:11.628000003Z" } > logcli query --forward --limit 100 -z UTC -q --no-labels --since 60m -o jsonl '{increment_duplicate_timestamp="true"}' \| jq { "line": "1a: 1683665111628000000", "timestamp": "2023-05-09T20:45:11.628Z" } { "line": "1b: 1683665111628000000", "timestamp": "2023-05-09T20:45:11.628000001Z" } { "line": "3: 1683665111628000003", "timestamp": "2023-05-09T20:45:11.628000003Z" } { "line": "2: 1683665111628000002", "timestamp": "2023-05-09T20:45:11.628000004Z" } > logcli query --forward --limit 100 -z UTC -q --no-labels --since 60m -o jsonl '{increment_duplicate_timestamp="false"}' \| jq { "line": "1a: 1683665111628000000", "timestamp": "2023-05-09T20:45:11.628Z" } { "line": "1b: 1683665111628000000", "timestamp": "2023-05-09T20:45:11.628Z" } { "line": "2: 1683665111628000002", "timestamp": "2023-05-09T20:45:11.628000002Z" } { "line": "3: 1683665111628000003", "timestamp": "2023-05-09T20:45:11.628000003Z" } ``` `true` and `false` were written to Loki 2.8.2 with the `-validation.increment-duplicate-timestamps` enabled or disabled `fixed` with this patch applied and the flag enabled Checklist - [x] Reviewed the [`CONTRIBUTING.md`](https://github.com/grafana/loki/blob/main/CONTRIBUTING.md) guide (required) - [ ] Documentation added - [x] Tests updated - [ ] `CHANGELOG.md` updated - [ ] Changes that require user attention or interaction to upgrade are documented in `docs/sources/upgrading/_index.md`	3 years ago
Dylan Guedes	1d20fdbf57	LogPushFailures: Log the tenant id (#9819 ) What this PR does / why we need it: Modify the `writefailures.Manager` to - Log the tenant id. Otherwise, it is hard to identify which tenant is responsible for a problematic push - Log stream parsing failures Which issue(s) this PR fixes: N/A	3 years ago
Dylan Guedes	dc5d458c14	Failures logging: avoid logging generic errors (#9796 ) What this PR does / why we need it: Modify our write path to not log generic messages with the `writefailures.Manager`. With generic messages I mean: contextCancel errors and contextDeadlineExceeded errors. Which issue(s) this PR fixes: N/A	3 years ago
Salva Corts	08e505dd50	Return all errors for a failed push request (#9807 ) What this PR does / why we need it: Distributors enforce limits and validate each push payload. before this PR, if a single push request yields multiple errors, only the last error is forwarded to the client. This PR modifies the distributor push handler so we forward all errors from a single push request to the client. Which issue(s) this PR fixes: Fixes #<issue number> Special notes for your reviewer: - I wonder if the use of [MultiError][1] is well-suited for this. [This test case][2] will return 100 errors with the same content. I wonder if we can leverage the MultiError to group errors that have the same content. Printing for example: `1 error like: 'Error parsing labels...'; 100 errors like: 'Max entry size...'`. This would require extra processing but If we do this processing in the Error() method I think it should be ok-ish? - Edit: I ended up implementing this in [`4b0816f`](`4b0816f67f`) Checklist - [ ] Reviewed the [`CONTRIBUTING.md`](https://github.com/grafana/loki/blob/main/CONTRIBUTING.md) guide (required) - [ ] Documentation added - [x] Tests updated - [ ] `CHANGELOG.md` updated - [ ] If the change is worth mentioning in the release notes, add `add-to-release-notes` label - [ ] Changes that require user attention or interaction to upgrade are documented in `docs/sources/upgrading/_index.md` - [ ] For Helm chart changes bump the Helm chart version in `production/helm/loki/Chart.yaml` and update `production/helm/loki/CHANGELOG.md` and `production/helm/loki/README.md`. [Example PR](`d10549e3ec`) [1]: `d4a4728933/pkg/util/errors.go (L32)` [2]: `f544b93c20/pkg/distributor/distributor_test.go (L75)`	3 years ago
Dylan Guedes	c0d9a2b4ba	Log push failures per-entry instead of in batches (#9720 ) What this PR does / why we need it: Modify how we log push errors to do it per-entry instead of accumulating all errors and then logging them all in a single pass. Which issue(s) this PR fixes: N/A	3 years ago
Trevor Whitney	4a56445686	Upgrade `golangci-lint` and fix linting errors (#9601 ) What this PR does / why we need it: Upgrade `golangci-lint` and fixes all the errors. The upgrade includes some stricter linting.	3 years ago
Travis Patterson	065bee7e72	Label Volume Endpoint (#9588 ) For a given set of matchers, returns the top N associated label/value pairs by volume. A query for `{cluster=prod}` will return ``` cluster=prod: size (total logs matching this matcher) . . . nth-label=nth-value ``` This is to service use cases where users want to understand where their log volume has come from by label without making multiple requests to the stats endpoint. Note: This PR is a monster but it's mostly plumbing. I've pointed out the most interesting bits that actually get the volumes from ingesters/indexs	3 years ago
Dylan Guedes	609bc22933	Distributor: Make key configurable when logging failures (#9659 ) What this PR does / why we need it: Make appending `insight=true` key-value pair to log failures configurable. Which issue(s) this PR fixes: N/A	3 years ago
Dylan Guedes	7d67b6306f	Loki: Implement initial phase of `limited_log_push_errors` (#9556 ) What this PR does / why we need it: Adds an initial implementation of `limited_log_push_errors`. This initial implementation encompass: - Runtime per-tenant configuration - Simple per-tenant rate-limiting based on error message size Notable features that will be added in future phases: - Instead of a single error with the final string, give to the manager the list of all entries - Once it supports per-entry error, make the rate-limiting separated per-reason - Hash the entry error and avoid repeating errors by caching seen errors in memory	3 years ago
Travis Patterson	fdc337692d	Stream Sharding amortize push payloads across the observed rate. (#9414 ) This should account for large payloads that come infrequently	3 years ago
cui fliter	520e434446	fix function names in comments (#9415 )	3 years ago
Trevor Whitney	50bc1cee71	Rename usagestats to analytics (#9212 ) What this PR does / why we need it: Rename usagestats to analytics. The config section in the yaml is already called analytics, this makes things more consistent. Checklist - [x] Reviewed the [`CONTRIBUTING.md`](https://github.com/grafana/loki/blob/main/CONTRIBUTING.md) guide (required) - [x] Documentation added - [ ] Tests updated - [x] `CHANGELOG.md` updated - [x] Changes that require user attention or interaction to upgrade are documented in `docs/sources/upgrading/_index.md`	3 years ago
Dylan Guedes	1bd5a20595	Distributor: Modify `createShard` to create slice with more precise size (#9265 ) What this PR does / why we need it: Modify the `createShard` method to create the entries slice with the number of entries divided by the total number of shards instead of always creating it with len=1024.	3 years ago
Salva Corts	eba255c945	Do not use IPv6 when getting addr from ring (#9255 ) What this PR does / why we need it: https://github.com/grafana/loki/pull/9184 changed the signature of `ring.GetInstanceAddr`. That function is now used since https://github.com/grafana/loki/pull/8067 which was not updated to use the changed signature. This PR fixes this by passing a new config flag `EnableIPv6` to the new `enableInet6` argument of `ring.GetInstanceAddr`. Similar to how it's done for the ruler: `3abd260975/pkg/ruler/base/ruler_ring.go (L80)` Which issue(s) this PR fixes: Fixes `main` Special notes for your reviewer: Checklist - [ ] Reviewed the [`CONTRIBUTING.md`](https://github.com/grafana/loki/blob/main/CONTRIBUTING.md) guide (required) - [ ] Documentation added - [ ] Tests updated - [ ] `CHANGELOG.md` updated - [ ] Changes that require user attention or interaction to upgrade are documented in `docs/sources/upgrading/_index.md` [1]: https://grafana.com/docs/loki/latest/configuration/	3 years ago
Dylan Guedes	24d426ce44	Distributor: Add AutoForget support (#8067 ) What this PR does / why we need it: - Add auto-forget capabilities to distributors - Change the distributor's ring implementation from Lifecycler to BasicLifecycler. That AutoForget delegate requires it.	3 years ago
Dylan Guedes	8580b8f8ad	Distributor: Reuse `time.Now()` for better performance (#9192 ) What this PR does / why we need it: Modifies `AggregateByShard` to reuse `time.Now()` to improve performance.	3 years ago
Dylan Guedes	bdfc0f92a0	Loki: Add debug event to ingester `GetStreamRates` (#9157 ) What this PR does / why we need it: - Add a new span event to our ingesters `GetStreamRates` function. - Add how many streams were updated on a `updateRates` call.	3 years ago
Dylan Guedes	b82ad11f92	Loki: Fix span event logging on rate store (#9086 ) What this PR does / why we need it: Fix an event logging on rate store that is missing the `event` key. `LogKV` expects a key-value pair always.	3 years ago
Dylan Guedes	9159c1dac3	Loki: Improve spans usage (#8927 ) What this PR does / why we need it: - At different places, inherit the span/spanlogger from the given context instead of instantiating a new one from scratch, which fix spans being orphaned on a read/write operation. - At different places, turn spans into events. Events are lighter than spans and by having fewer spans in the trace, trace visualization will be cleaner without losing any details. - Adds new spans/events to places that might be a bottleneck for our writes/reads.	3 years ago
Danny Kopping	d38d481f35	Distributor: add detail to stream rates failure (#8900 ) What this PR does / why we need it: Currently we cannot see if a single ingester is the source of the `unable to get stream rates` failures. This change adds the client address to the log entry. Signed-off-by: Danny Kopping <danny.kopping@grafana.com>	3 years ago
Dylan Guedes	4f94b89fdb	Loki: Add more spans to write path (#8888 ) What this PR does / why we need it: Add new spans to our write path to better determine which operations are holding our write performance.	3 years ago
Callum Styan	9a2a038f43	Allow passing of context to query related limits functions (#8689 ) In this PR we're allowing for passing of a `context.Context` via the Limits interfaces (some of which are new, to clean up hardcoding/embedding of `validation.Overrides`) This is based on work/ideas by @jeschkies . Fixes #8694 --------- Signed-off-by: Callum Styan <callumstyan@gmail.com> Co-authored-by: Karsten Jeschkies <karsten.jeschkies@grafana.com>	3 years ago
zhuoyuan-liu	571f88bc57	Reduce distributor code duplication (#8293 ) Co-authored-by: Michel Hollands <42814411+MichelHollands@users.noreply.github.com>	3 years ago
Kaviraj Kanagaraj	db3a9c961e	distributor: Fix `loki_mutated_bytes_total` metric increment (#8223 ) Signed-off-by: Kaviraj <kavirajkanagaraj@gmail.com> Previously this metric was not not incremented for all the log lines in the stream. This PR fixes it.	3 years ago
Dylan Guedes	2e496ddff2	Loki: Add one additional bucket to `rate_store_stream_shards` (#7856 ) What this PR does / why we need it: Add one additional bucket (value=1) to the `rate_store_stream_shards`. By jumping from 0 to 2 we can't distinguish between streams that were minimally sharded (shards=2) to non-sharded ones (shards=1).	3 years ago
Travis Patterson	a1e9e4cdd0	Revert "Make some stream shard metrics per-tenant (#7838 )" (#7842 ) Because these metrics will be coming from each distributor, this is going to have an adverse impact on cardinality. Consider doing this with logging	3 years ago
Travis Patterson	96fc2abfff	Make some stream shard metrics per-tenant (#7838 ) This PR just adds tenant information so some of our per-stream-rate-limit metrics.	3 years ago
Travis Patterson	eb7fbfe0b4	Improve ratestore performance (#7810 ) This PR changes the Rate Store to use multi-level maps rather than `tenant+stream-fingerprint` key. Benchmarks: old ``` goos: linux goarch: amd64 pkg: github.com/grafana/loki/pkg/distributor cpu: 11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz BenchmarkRateStore-8 117 10420080 ns/op 3201761 B/op 200025 allocs/op PASS ok github.com/grafana/loki/pkg/distributor 14.646s ``` new ``` goos: linux goarch: amd64 pkg: github.com/grafana/loki/pkg/distributor cpu: 11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz BenchmarkRateStore-8 320 3811906 ns/op 2660 B/op 26 allocs/op ``` stats ``` name old time/op new time/op delta RateStore-8 10.7ms ± 8% 4.1ms ± 8% -61.42% (p=0.000 n=10+10) name old alloc/op new alloc/op delta RateStore-8 3.20MB ± 0% 0.00MB ± 0% -99.92% (p=0.000 n=8+9) name old allocs/op new allocs/op delta RateStore-8 200k ± 0% 0k ± 0% -99.99% (p=0.000 n=9+10) ```	3 years ago
Dylan Guedes	4567d01ac6	Loki: Add new sharding metrics (histogram based) (#7762 ) What this PR does / why we need it: Add histogram-version of two existing stream sharding metrics. The new metrics are: - `rate_store_stream_shards`: Will show us the distribution of shards reported by ingesters. This will help us on monitoring the number of shards we're dealing with. As of now we only have access to the max number of shards seen, which is helpful but not enough. - `rate_store_stream_rate_bytes`: Same as below: will help us understand the distribution of stream sizes instead of only showing us the max stream size seen.	3 years ago
Travis Patterson	af68dbc6cf	Keep rates in the ratestore for 10 minutes (#7599 ) This PR accounts for intermittent streams that might get dropped on a per-second update interval Co-authored-by: Dylan Guedes <djmgguedes@gmail.com>	3 years ago
Travis Patterson	c75b822fc6	Track rates per stream and per tenant (#7564 ) It is possible for two tenants to have streams with the same labels. This PR changes the stream tracking in the `shard_tracker`, `stream_rate_calculator`, and `rate_store` to account for tenant ID when tracking streams.	3 years ago
Travis Patterson	a65da7d78e	Shard streams round-robin rather in batches. (#7560 ) To avoid hot spots, keep track of shard numbers across pushes. When tracking shard numbers across pushes, it's no longer necessary to randomize shards, in the case where the desired number of shards is greater than the number of entries. Instead, we just partition the entries among the shards in round robin fashion. Over time, this should give us a more even distribution of logs in shards, particularly for large streams that are made of small individual pushes.	3 years ago
Travis Patterson	0a654e2357	Ensure labels on sharded streams are sorted before they're sent anywhere (#7556 ) Pushing streams with unsorted labels to ingesters results in a bug where ingesters duplicate chunks for streams. This causes a memory leak and eventual OOM crash when ingesters recover from the WAL because one of the duplicate streams is orphaned. This PR ensures that sharded streams have sorted labels to avoid this bug.	3 years ago
Callum Styan	46b40a5582	Hardcode ring.WriteNoExtend for distributor push operations (#7517 ) With zone awareness enabled we allow restarting multiple ingesters within a single zone at the same time, this means that for many more write requests there's potentially only 2 of the 3 replicas for a stream up at the time the write request is received. This lead to a large increase in 500s during rollouts. My suspicion is that the 500s on rollouts was always related to the following; the replica set ring code has something referred to as "extending the replica set". This essentially is the adding of ingesters in the ring that are not in an ACTIVE state (for example, are in LEAVING) to the set of replicas considered valid for an operation. Adding of the instance in a state other than ACTIVE can be controlled by using a different operation type, which Mimir was using but we were not. Here are the relevant functions, [ring.Get](`8d6d914ef6/ring/ring.go (L335-L403)`) (particularly these lines [1](`8d6d914ef6/ring/ring.go (L381-L389)`) and [2](`8d6d914ef6/ring/ring.go (L394-L397)`)) and [replicationStrategy.Filter](`789ec0ca4a/ring/replication_strategy.go (L22-L71)`) (particularly these lines [1](`789ec0ca4a/ring/replication_strategy.go (L32-L36)`) and [2](`789ec0ca4a/ring/replication_strategy.go (L54-L67)`)) The tl; dr of the above code snippets is that we seem to always have required a minSuccess from replicationFactor/2 + 1 except we overwrite replicationFactor to be the # of instances which is likely always going to be more than just 3 ingester replicas because of the replica set extension. At least for zone aware rollouts this seems to be the case. This is because many ingesters will be in the LEAVING state. We can avoid this by just using `WriteNoExtend` operations instead of `Write` operations ([here](`789ec0ca4a/ring/ring.go (L83-L93)`)) In this PR I've hardcoded our call to the ring in `distributor.Push` to always use `ring.WriteNoExtend`. IMO this is the better option to allowing `ring.WriteNoExtend` to be used optionally and defaulting to `ring.Write`. Extending the replication set on writes feels like a footgun for write outages and there's even some feelings internally that allowing that extension is a bug. For more background information see: https://github.com/grafana/mimir/issues/1854 Signed-off-by: Callum Styan <callumstyan@gmail.com> Signed-off-by: Callum Styan <callumstyan@gmail.com>	3 years ago
Cyril Tovena	7b01864a91	Improves performance of distributor validations (#7534 ) Most of the time we don't need to create those strings which create an alloc per entry.	3 years ago
Travis Patterson	45d59ba571	Improve rate store refresh latency (#7492 ) Rather than iterating streams and getting rates per stream, keep all necessary information in the rate calculator and just hand back the current rates when asked. This change took p99 latencies from ~1.8s to ~400ms and p50 latences from ~1s to 6ms.	3 years ago
Dylan Guedes	28e192f22e	Loki: Implement random entries assignment when sharding (#7427 ) What this PR does / why we need it: Modify sharding to assign random shards when the number of shards is bigger than the number of entries. This helps us to avoid hot-spotting the first shards. By randomizing the entries assignment we help spread the load between all shards.	3 years ago
Travis Patterson	8c84ee8b7a	Fix stream sharding (#7393 ) Promtail doesn't add a stream hash to push requests. As a result, we weren't accounting for the current rate in our sharding calculation and only sharding on individual large pushes. This PR ensures that streams have a hash as they're ingested. It also add a couple of metrics that were useful in debugging this issue.	3 years ago
Travis Patterson	543ea78b7a	Add metrics to the distributor about stream sharding (#7390 ) This PR adds some more metrics to the steam sharding interactions from the Distributor. 1. `rate_store_stream_count`: The total number of streams 2. `rate_store_max_stream_shard_count`: The stream with the largest number of shards 3. `rate_store_max_stream_rate_bytes`: The largest stream seen by a distributor 4. `rate_store_refresh_duration_seconds`: A histogram of the latency to query all the ingesters and process their rates	3 years ago
Travis Patterson	5dc27e8064	Fix Flaky Distributor Test (#7391 ) Use eventually for async tests	3 years ago
Travis Patterson	3e7adb2c34	Measure Per-Stream rates (#7370 ) This PR implements the ability for ingesters to measure per-stream rates at 1/s resolution. It also fixes bugs in the kind of Limits used in the RateStore and the Ingester API	3 years ago
Travis Patterson	7979cfbe07	Implement Distributor Rate Store (#7349 ) The `RateStore` concurrently queries all ingester `GetStreamRate` APIs and stores the results. `RateStore` deduplicates replicas and combines sharded streams so a caller of `RateFor` gets an unsharded view of a stream's rate.	3 years ago
Dylan Guedes	b1d4efab12	Loki: Per-tenant stream sharding (#7311 ) What this PR does / why we need it: - Move stream sharding configuration to its own package to avoid cyclic imports - Change stream sharding to be a per-tenant configuration - Change ingesters to reject whole streams due to rate-limit based on per-tenant stream sharding - Change stream sharding flags prefix from `distributor.shard-stream` to `shard-stream`	3 years ago
Travis Patterson	e3e4e27c8c	Implement ingester API to return stream rates (#7176 ) This PR introduces an API on the Ingester that reports all current streams in memory and their current rate in bytes/second. The reponse from the API contains a slice of `StreamRate` structs: ```go type StreamRate struct { StreamHash uint64 `protobuf:"varint,1,opt,name=streamHash,proto3" json:"streamHash,omitempty"` StreamHashNoShard uint64 `protobuf:"varint,2,opt,name=streamHashNoShard,proto3" json:"streamHashNoShard,omitempty"` Rate int64 `protobuf:"varint,3,opt,name=rate,proto3" json:"rate,omitempty"` } ``` In the interest of space on the caller and in transmission, only hashes and rates are sent. - `StreamHash` will be used to deduplicate replicas - `StreamHashNoShard` will be used to aggregate a sharded stream's total rate. Co-authored-by: Dylan Guedes <djmgguedes@gmail.com>	3 years ago
Dylan Guedes	da7acb49e5	Loki: Add sharding support for negative/zeroed desired rate (#7342 ) What this PR does / why we need it: Adds the handling of negative/zeroed desired rate to our distributor stream sharding. Without this, invalid desired rates results in a -Inf number of shards. This also adds a log line to report that such a thing happened. Signed-off-by: Danny Kopping <danny.kopping@grafana.com> Co-authored-by: Danny Kopping <danny.kopping@grafana.com>	3 years ago

1 2 3 4

190 Commits (89985399817dde28fe5b94de990133d904bc3e5b)