📅 0001年01月01日 · ☕ 1 分钟

tcmalloc，通过-define tcmalloc=disabled禁用

https://blog.gmem.cc/envoy-study-note

📅 0001年01月01日 · ☕ 1 分钟

https://istio.cn/t/topic/299

Envoy源码分析之Dispatcher：https://developer.aliyun.com/article/659277

线程相关 Misc：

Envoy进程由一个Main Thread和多个Worker Thread 组成
每个Main和Worker包含一个eventloop，所有的处理都是由eventloop触发开始
Main负责xDS等功能，Worker负责处理连接和请求
当一个client向Envoy建立连接的时候，因为所有Worker的EventLoop都注册了listening fd(启用SO_PORTREUSE除外)，会由内核决定分配给哪个Worker
当一个下游client连接到了Envoy，在保持连接不断的情况下，会和同一个Worker进行通讯

📅 0001年01月01日 · ☕ 2 分钟

HTTP/1.1 Header Casing

https://www.envoyproxy.io/docs/envoy/latest/configuration/http/http_conn_man/header_casing#config-http-conn-man-header-casing

When handling HTTP/1.1, Envoy will normalize the header keys to be all lowercase. While this is compliant with the HTTP/1.1 spec, in practice this can result in issues when migrating existing systems that might rely on specific header casing.

To support these use cases, Envoy allows configuring a formatting scheme for the headers, which will have Envoy transform the header keys during serialization.

To configure this formatting on response headers, specify the format in the http_protocol_options.
To configure this for upstream request headers, specify the formatting in http_protocol_options in the cluster’s extension_protocol_options.

Currently Envoy supports two mutually exclusive types of header key formatters:

📅 0001年01月01日 · ☕ 2 分钟

config.core.v3.Http1ProtocolOptions

https://www.envoyproxy.io/docs/envoy/latest/api-v3/config/core/v3/protocol.proto#config-core-v3-http1protocoloptions

[config.core.v3.Http1ProtocolOptions proto]

1
2
3
4
5
6
7
8
9


{
  "allow_absolute_url": "{...}",
  "accept_http_10": "...",
  "default_host_for_http_10": "...",
  "header_key_format": "{...}",
  "enable_trailers": "...",
  "allow_chunked_length": "...",
  "override_stream_error_on_invalid_http_message": "{...}"
}

allow_absolute_url

(BoolValue) Handle HTTP requests with absolute URLs in the requests. These requests are generally sent by clients to forward/explicit proxies. This allows clients to configure envoy as their HTTP proxy. In Unix, for example, this is typically done by setting the http_proxy environment variable.

📅 0001年01月01日 · ☕ 1 分钟

Enable debug log by command line

https://projectcontour.io/docs/v1.10.0/troubleshooting/envoy-debug-log/

The envoy command has a --log-level flag that can be useful for debugging. By default, it’s set to info. To change it to debug, edit the envoy DaemonSet in the projectcontour namespace and replace the --log-level info flag with --log-level debug. Setting the Envoy log level to debug can be particilarly useful for debugging TLS connection failures.

Enable debug log by API

列出 logger 名字：

📅 0001年01月01日 · ☕ 2 分钟

https://www.envoyproxy.io/docs/envoy/latest/operations/performance

Performance¶

Envoy is architected to optimize scalability and resource utilization by running an event loop on a small number of threads. The “main” thread is responsible for control plane processing, and each “worker” thread handles a portion of the data plane processing. Envoy exposes two statistics to monitor performance of the event loops on all these threads.

Loop duration: Some amount of processing is done on each iteration of the event loop. This amount will naturally vary with changes in load. However, if one or more threads have an unusually long-tailed loop duration, it may indicate a performance issue. For example, work might not be distributed fairly across the worker threads, or there may be a long blocking operation in an extension that’s impeding progress.

📅 0001年01月01日 · ☕ 3 分钟

Kernel

https://lwn.net/Articles/542629/
One of the features merged in the 3.9 development cycle was TCP and UDP support for the SO_REUSEPORT socket option; that support was implemented in a series of patches by Tom Herbert. The new socket option allows multiple sockets on the same host to bind to the same port, and is intended to improve the performance of multithreaded network server applications running on top of multicore systems.
The basic concept of SO_REUSEPORT is simple enough. Multiple servers (processes or threads) can bind to the same port if they each set the option as follows:

📅 0001年01月01日 · ☕ 2 分钟

https://www.envoyproxy.io/docs/envoy/latest/intro/arch_overview/upstream/connection_pooling#automatic-protocol-selection

Connection pooling

For HTTP traffic, Envoy supports abstract connection pools that are layered on top of the underlying wire protocol (HTTP/1.1, HTTP/2, HTTP/3). The utilizing filter code does not need to be aware of whether the underlying protocol supports true multiplexing or not. In practice the underlying implementations have the following high level properties:

HTTP/1.1

The HTTP/1.1 connection pool acquires connections as needed to an upstream host (up to the circuit breaking limit). Requests are bound to connections as they become available, either because a connection is done processing a previous request or because a new connection is ready to receive its first request. The HTTP/1.1 connection pool does not make use of pipelining so that only a single downstream request must be reset if the upstream connection is severed.

📅 0001年01月01日 · ☕ 2 分钟

TLS ALPN

Force HTTP 1.1

https://www.tetrate.io/blog/envoy-and-istio-security-releases-june-2020/

Envoy versions can mitigate those vulnerabilities by disabling HTTP2 and allowing only HTTP/1.1 by setting http_connection_manager.codec_type to “HTTP1” and removing “h2” from common_tls_context.alpn_protocols.

For Istio:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23


apiVersion: networking.istio.io/v1alpha3
kind: EnvoyFilter
metadata:
  name: disable-ingress-h2
  namespace: istio-system
spec:
  workloadSelector:
    labels:
      istio: ingressgateway
  configPatches:
  - applyTo: NETWORK_FILTER # http connection manager is a filter in Envoy
    match:
      context: GATEWAY
      listener:
        filterChain:
          filter:
            name: "envoy.http_connection_manager"
    patch:
      operation: MERGE
      value:
        typed_config:
          "@type": type.googleapis.com/envoy.config.filter.network.http_connection_manager.v2.HttpConnectionManager
          codec_type: HTTP1

HTTP 1.1 Upgrade header

https://www.envoyproxy.io/docs/envoy/latest/intro/arch_overview/http/upgrades#http-upgrades

📅 0001年01月01日 · ☕ 3 分钟

HTTP Protocol Options

https://www.envoyproxy.io/docs/envoy/latest/api-v3/extensions/upstreams/http/v3/http_protocol_options.proto#extensions-upstreams-http-v3-httpprotocoloptions-autohttpconfig

This extension may be referenced by the qualified name envoy.upstreams.http.http_protocol_options

extensions.upstreams.http.v3.HttpProtocolOptions

HttpProtocolOptions specifies Http upstream protocol options. This object is used in typed_extension_protocol_options, keyed by the name envoy.extensions.upstreams.http.v3.HttpProtocolOptions.

This controls what protocol(s) should be used for upstream and how said protocol(s) are configured.

This replaces the prior pattern of explicit protocol configuration directly in the cluster. So a configuration like this, explicitly configuring the use of HTTP/2 upstream:

📅 0001年01月01日 · ☕ 0 分钟

📅 0001年01月01日 · ☕ 8 分钟

https://helm.sh/docs/topics/charts/

Chart

A Chart is a Helm package. It contains all of the resource definitions necessary to run an application, tool, or service inside of a Kubernetes cluster. Think of it like the Kubernetes equivalent of a Homebrew formula, an Apt dpkg, or a Yum RPM file.

Directory structure

wordpress/
  Chart.yaml          # A YAML file containing information about the chart
  LICENSE             # OPTIONAL: A plain text file containing the license for the chart
  README.md           # OPTIONAL: A human-readable README file
  values.yaml         # The default configuration values for this chart
  values.schema.json  # OPTIONAL: A JSON Schema for imposing a structure on the values.yaml file
  charts/             # A directory containing any charts upon which this chart depends.
  crds/               # Custom Resource Definitions
  templates/          # A directory of templates that, when combined with values,
                      # will generate valid Kubernetes manifest files.
  templates/NOTES.txt # OPTIONAL: A plain text file containing short usage notes

Pull:

📅 0001年01月01日 · ☕ 1 分钟

As you edit your chart, you can validate that it is well-formed by running helm lint.

When it’s time to package the chart up for distribution, you can run the helm package command:

1
2


$ helm package deis-workflow
deis-workflow-0.1.0.tgz

Chat

HTTP/1.1 Header Casing

config.core.v3.Http1ProtocolOptions

Enable debug log by command line

Enable debug log by API

Performance¶

Kernel

Connection pooling

HTTP/1.1

TLS ALPN

Force HTTP 1.1

HTTP 1.1 Upgrade header

HTTP Protocol Options

extensions.upstreams.http.v3.HttpProtocolOptions

Chart

Directory structure