Hello!
I'm Marco, a Software Engineer at Grafana Labs.
I love challenges, building products, distributed systems and all things observability. I'm a Grafana Mimir maintainer.
pracucci
Follow
Pinned
5,112 contributions in the last year
We're celebrating 100 million developers!
Activity overview
Contribution activity
January 2023
Created 34 commits in 5 repositories
Created a pull request in prometheus/prometheus that received 10 comments
API: change HTTP status code from 503/422 to 499 if a request is canceled
Grafana Mimir uses the Prometheus API. I noticed the query API returns 503 when the request is canceled. As a side effect, this causes canceled que…
+13
−3
•
10
comments
Opened 20 other pull requests in 2 repositories
grafana/mimir
1
open
16
merged
- Upgrade upstream Prometheus: add keep_firing_for support to alerting rules and track 499 on canceled requests
- Fix TestBucketStore_Series_LimitsWithStreamingEnabled
- Fix MimirAutoscalerNotActive alert to not fire if the scaling metric does not exist and introduce MimirAutoscalerKedaFailing alert
- Store-gateway: apply limits on merged series
- Remove unused resolution, step and range parameters from store-gateway SeriesRequest
- Mark query-scheduler ring-based service discovery as stable and document migration
- Add --concurrency support to 'mimirtool rules sync' command
- Improve streaming store-gateway metrics
- Announce next Mimir release shepherd
- Fix mimir-read ingress service
- Changed default mimir_backend_data_disk_size from 100Gi to 250Gi
- Jsonnet: replace -store.max-query-length with -query-frontend.max-total-query-length
- Fix mimir-read CLI flags precedence
- Fix mimir-read Kubernetes service
- Fix ruler-query-frontend and mimir-read gRPC server configuration
- Fix mimir-write Kubernetes service to correctly balance route requests among mimir-write pods
- Chore: removed unused ingester's ingesterClientFactory
grafana/mimir-prometheus
2
merged
1
closed
Reviewed 136 pull requests in 8 repositories
grafana/mimir
25 pull requests
- Streaming store-gateway: chunks grouping
- Upgrade upstream Prometheus: add keep_firing_for support to alerting rules and track 499 on canceled requests
- [Docs] Clarify how ingester replication happens
- Clarify that memberlist cannot be used for HA dedupe.
- store-gateway: add partitioners for each data type
- Querier: propagate errors with status code 422 coming from the store-gateway
- Make use of a per-call memory allocator for loading cached chunks
- Include user agent in query stats logs
- Replace short flag names with long flag names in the runbook for better clarity
- Update doc-validator workflow so that it no longer relies on old checkout mechanism
- Migrate fpm to nfpm
-
Run
ExpandedPostingstest and benchmark with both streaming and mmap-based index header reader. - Mixin: Add gateway auto-scaling panels to reads and writes dashboards
- query-frontend: During shut-down, wait on in-flight queries
- Docs: Explain how to run Mimir locally using docker-compose
- Address Marco's feedback on 4055
- Optimize prefix matching
- Add Histogram to mimir.proto
- WIP - Update dskit with cache refactor changes
- Vendor update mimir-prometheus
- Deduplicate ring config code
- Enforce max_fetched_series_per_query in store-gateway
-
Return
Canceledwhen aLabelNamesorLabelValuesrequest to a store-gateway is cancelled by the caller, and returnInternalotherwise for all requests. - Query-frontend: Disambiguate logs for failed queries.
- Distinguish between label names and label values
- Some pull request reviews not shown.
grafana/dskit
7 pull requests
- Fix: Readme included cortex-related references
- Add Delete method to cache.Cache interface
- multierror: Add support for errors.Is()
- spanlogger: skip formatting if trace is not sampled
- Introduce support for per-call memory allocator for caching
- Use Grafana Memcached client fork instead of upstream
- Remove trivial "util" exported package
grafana/mimir-prometheus
4 pull requests
prometheus/prometheus
2 pull requests
grafana/rollout-operator
2 pull requests
thanos-io/objstore
1 pull request
grafana/gomemcache
1 pull request
grafana/e2e
1 pull request
Created an issue in grafana/mimir that received 3 comments
Querier tracks cortex_request_duration_seconds_count tracks status_code=503 on "context canceled"
I'm investigating the alert MimirRequestErrors which fired for a short period on a Mimir cluster because the querier tracked cortex_request_duratio…
3
comments
Opened 6 other issues in 1 repository
grafana/mimir
4
closed
2
open
- Docs: use long flag names in runbook commands
- Store-gateway: Series() response latency impacted by Send() more than I expected
- mimirtool rules sync should parallelise the API calls
- Store-gateway: investigate and eventually improve preloading efficiency
- Store-gateway: inefficient chunks fetching and caching
- Possibly unbalanced gRPC load balancing from remote ruler
Answered 4 discussions in 1 repository
grafana/mimir
grafana/mimir
-
at least 2 live replicas required across different availability zones, could only find 1
This contribution was made on Jan 20
-
mimir Monolithic - 2 VM cluster - err-mimir-bucket-index-too-old
This contribution was made on Jan 3
-
mimir Monolithic - 2 VM cluster - err-mimir-bucket-index-too-old
This contribution was made on Jan 3
-
mimir Monolithic - 2 VM cluster - err-mimir-bucket-index-too-old
This contribution was made on Jan 3
233
contributions
in private repositories
Jan 2 – Jan 30