服务监控系统 Prometheus 2.4.0 已发布，该版本包含多项修正和改进。由于 WAL 实现已被重写，因此存储不向前兼容。

Prometheus 是一个 Go 语言开发的开源的服务监控系统和时间序列数据库。

更新内容：

• [更改] Reduce remote write default retries #4279

• [更改] Remove /heap endpoint #4460

• [特性] Persist alert 'for' state across restarts #4061

• [特性] Add API providing per target metric metadata #4183

• [特性] Add API providing recording and alerting rules #4318 #4501

• [改进] Brand new WAL implementation for TSDB. Forwards incompatible with previous WAL.

• [改进] Show rule evaluation errors in UI #4457

• [改进] Throttle resends of alerts to Alertmanager #4538

• [改进] Send EndsAt along with the alert to Alertmanager #4550

• [改进] Limit the samples returned by remote read endpoint #4532

• [改进] Limit the data read in through remote read #4239

• [改进] Coalesce identical SD configuations #3912

• [改进] promtool: Add new commands for debugging and querying #4247 #4308 #4346#4454

• [改进] Update console examples for node_exporter v0.16.0 #4208

• [改进] Optimize PromQL aggregations #4248

• [改进] Remote read: Add Offset to hints #4226

• [改进] consul_sd: Add support for ServiceMeta field #4280

• [改进] ec2_sd: Maintain order of subnet_id label #4405

• [改进] ec2_sd: Add support for custom endpoint to support EC2 compliant APIs #4333

• [改进] ec2_sd: Add instance_owner label #4514

• [改进] azure_sd: Add support for VMSS discovery and multiple environments #4202#4569

• [改进] gce_sd: Add instance_id label #4488

• [改进] Forbid rule-abiding robots from indexing #4266

• [改进] Log virtual memory limits on startup #4418

• [修复] Wait for service discovery to stop before exiting #4508

• [修复] Render SD configs properly #4338

• [修复] Only add LookbackDelta to vector selectors #4399

• [修复] ec2_sd: Handle panic-ing nil pointer #4469

• [修复] consul_sd: Stop leaking connections #4443

• [修复] Use templated labels also to identify alerts #4500

• [修复] Reduce floating point errors in stddev and related functions #4533

• [修复] Log errors while encoding responses #4359

下载地址：

• https://prometheus.io/download/