mirror of
https://github.com/VictoriaMetrics/VictoriaMetrics.git
synced 2024-12-03 16:21:14 +01:00
lib/netutil.ConnPool: skip dialing remote address if the previous dial attempt was unsuccessful
If the previous dial attempt was unsuccessful, then all the new dial attempts are skipped until the background goroutine determines that the given address can be successfully dialed. This reduces query latency when some of vmstorage nodes are unavailable and dialing them is slow. This should help with https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711 This commit is based on ideas from the https://github.com/VictoriaMetrics/VictoriaMetrics/pull/2756 The main differences are: - The check for healthy/unhealthy storage nodes is moved one level lower from app/vmselect/netstorage to lib/netutil.ConnPool. This makes possible re-using this feature everywhere lib/netutil.ConnPool is used. - The check doesn't take into account handshake errors for already established connections. Handshake errors usually mean improperly configured VictoriaMetrics cluster, so they shouldn't be ignored.
This commit is contained in:
parent
49586566a3
commit
3837b50f37
@ -39,6 +39,8 @@ The following tip changes can be tested by building VictoriaMetrics components f
|
||||
* FEATURE: [vmagent](https://docs.victoriametrics.com/vmagent.html): add `-promscrape.cluster.name` command-line flag, which allows proper data de-duplication when the same target is scraped from multiple [vmagent clusters](https://docs.victoriametrics.com/vmagent.html#scraping-big-number-of-targets). See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2679).
|
||||
* FEATURE: [vmagent](https://docs.victoriametrics.com/vmagent.html): add `action: graphite` relabeling rules optimized for extracting labels from Graphite-style metric names. See [these docs](https://docs.victoriametrics.com/vmagent.html#graphite-relabeling) and [this feature request](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2737).
|
||||
* FEATURE: [VictoriaMetrics enterprise](https://victoriametrics.com/products/enterprise/): expose `vm_downsampling_partitions_scheduled` and `vm_downsampling_partitions_scheduled_size_bytes` metrics, which can be used for tracking the progress of initial [downsampling](https://docs.victoriametrics.com/#downsampling) for historical data. See [this feature request](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2612).
|
||||
* FEATURE: [VictoriaMetrics cluster](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html): do not spend up to 5 seconds when trying to connect to unavailable `vmstorage` nodes. This should improve query latency when some of `vmstorage` nodes aren't available. Expose `vm_tcpdialer_addr_available{addr="..."}` metric at `http://vmselect:8481/metrics` for determining whether the given `addr` is available for establishing new connections. See [this comment](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711#issuecomment-1160363187).
|
||||
* FEATURE: [VictoriaMetrics cluster](https://docs.victoriametrics.com/Cluster-VictoriaMetrics.html): add `-vmstorageDialTimeout` command-line flags to `vmselect` and `vminsert` for tuning the maximum duration for connection estabilishing to `vmstorage` nodes. This should help resolving [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711).
|
||||
|
||||
* BUGFIX: support for data ingestion in [DataDog format](https://docs.victoriametrics.com/Single-server-VictoriaMetrics.html#how-to-send-data-from-datadog-agent) from legacy clients / agents. See [this pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/2670). Thanks to @elProxy for the fix.
|
||||
* BUGFIX: [vmagent](https://docs.victoriametrics.com/vmagent.html): do not expose `vm_promscrape_service_discovery_duration_seconds_bucket` metric for unused service discovery types. This reduces the number of metrics exported at `http://vmagent:8429/metrics`. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2671).
|
||||
|
Loading…
Reference in New Issue
Block a user