Varnish 3.x

Description

Varnish is a caching front-end server. This document has notes on how to use Varnish with Plone. If you're using Varnish 4.x, then you want to look into Varnish 4.x

Introduction

This chapter contains information about using the Varnish caching proxy with Plone.

To use Varnish with Plone

  • Learn how to install and configure Varnish
  • Add Plone virtual hosting rule to the default varnish configuration

Note

Some of these examples were written for Varnish 2.x. Varnish 3.x (Jun 2011) has radically altered syntax of VCL language and command line tools, so you might need to adapt the examples a bit.

Installation

The suggest method to install Varnish is to use your OS package manager.

  • You can install using packages (RPM/DEB) - consult your operating system instructions.
  • For more up to date packages for Debian you could check: https://www.varnish-cache.org/installation/debian
  • You can install backports
  • You can install using buildout

Buildout examples

Management console

varnishadm

You can access Varnish admin console on your server by:

# Your system uses a secret handshake file
varnishadm -T localhost:6082 -S /etc/varnish/secret

(Ubuntu/Debian installation)

Telnet console

The telnet management console is available on some configurations where varnishadm cannot be used. The functionality is the same.

Example:

ssh yourhost
# Your system does not have a secret handshake file
telnet localhost 6082

Note

Port number depends on your Varnish settings.

More info

Quit console

Quit command:

quit

Purging the cache

This will remove all entries from the Varnish cache:

varnishadm "ban.url ."

Loading new VCL to live Varnish

More often than not, it is beneficial to load new configuration without bringing the cache down for maintenance. Using this method also checks the new VCL for syntax errors before activating it. Logging in to Varnish CLI requires the varnishadm tool, the address of the management interface, and the secret file for authentication.

See the varnishadm man-page for details.

Opening a new CLI connection to the Varnish console, in a buildout-based Varnish installation:

parts/varnish-build/bin/varnishadm -T localhost:8088

Port 8088 is defined in buildout.cfg:

[varnish-instance]
telnet = localhost:8088

Opening a new CLI connection to the Varnish console, in a system-wide Varnish installation on Ubuntu/Debian:

varnishadm -T localhost:6082 -S /etc/varnish/secret

You can dynamically load and parse a new VCL config file to memory:

vcl.load <name> <file>

For example:

vcl.load newconf_1 /etc/varnish/newconf.vcl

... or ...

# Ubuntu / Debian default config
vcl.load defconf1 /etc/varnish/default.vcl

vcl.load will load and compile the new configuration. Compilation will fail and report on syntax errors. Now that the new configuration has been loaded, it can be activated with:

vcl.use newconf_1

Note

Varnish remembers <name> in vcl.load, so every time you need to reload your config you need to invent a new name for vcl.load / vcl.use command pair.

See

Logs

To see a real-time log dump (in a system-wide Varnish configuration):

varnishlog

By default, Varnish does not log to any file and keeps the log only in memory. If you want to extract Apache-like logs from varnish, you need to use the varnishncsa utility.

Stats

Check live "top-like" Varnish statistics:

parts/varnish-build/bin/varnishstat

Use the admin console to print stats for you:

stats
200 2114

       95717  Client connections accepted
      132889  Client requests received
       38638  Cache hits
       21261  Cache hits for pass
      ...

Virtual hosting proxy rule

Varnish 3.x example

An example with two separate Plone installations (Zope standalone mode) behind Varnish 3.x HTTP 80 port.

Example:

#
# This backend never responds... we get hit in the case of bad virtualhost name
#
backend default {
    .host = "127.0.0.1";
    .port = "55555";
}

#
# Plone Zope front end clients running on koskela
#
backend site1 {
    .host = "127.0.0.1";
    .port = "9944";
}

backend site2 {
    .host = "127.0.0.1";
    .port = "9966";
}

#
# Guess which site / virtualhost we are diving into.
# Apache, Nginx or Plone directly
#
sub choose_backend {

    if (req.http.host ~ "^(.*\.)?site2\.fi(:[0-9]+)?$") {
        set req.backend = site2;

        # Zope VirtualHostMonster
        set req.url = "/VirtualHostBase/http/" + req.http.host + ":80/Plone/VirtualHostRoot" + req.url;

    }

    if (req.http.host ~ "^(.*\.)?site1\.fi(:[0-9]+)?$") {
        set req.backend = site1;

        # Zope VirtualHostMonster
        set req.url = "/VirtualHostBase/http/" + req.http.host + ":80/Plone/VirtualHostRoot" + req.url;
    }

}


sub vcl_recv {

    #
    # Do Plone cookie sanitization, so cookies do not destroy cacheable anonymous pages
    #
    if (req.http.Cookie) {
        set req.http.Cookie = ";" + req.http.Cookie;
        set req.http.Cookie = regsuball(req.http.Cookie, "; +", ";");
        set req.http.Cookie = regsuball(req.http.Cookie, ";(statusmessages|__ac|_ZopeId|__cp)=", "; \1=");
        set req.http.Cookie = regsuball(req.http.Cookie, ";[^ ][^;]*", "");
        set req.http.Cookie = regsuball(req.http.Cookie, "^[; ]+|[; ]+$", "");

        if (req.http.Cookie == "") {
            remove req.http.Cookie;
        }
    }

    call choose_backend;

    if (req.request != "GET" &&
      req.request != "HEAD" &&
      req.request != "PUT" &&
      req.request != "POST" &&
      req.request != "TRACE" &&
      req.request != "OPTIONS" &&
      req.request != "DELETE") {
        /* Non-RFC2616 or CONNECT which is weird. */
        return (pipe);
    }
    if (req.request != "GET" && req.request != "HEAD") {
        /* We only deal with GET and HEAD by default */
        return (pass);
    }
    if (req.http.Authorization || req.http.Cookie) {
        /* Not cacheable by default */
        return (pass);
    }
    return (lookup);
}


#
# Show custom helpful 500 page when the upstream does not respond
#
sub vcl_error {
  // Let's deliver a friendlier error page.
  // You can customize this as you wish.
  set obj.http.Content-Type = "text/html; charset=utf-8";
  synthetic {"
  <?xml version="1.0" encoding="utf-8"?>
  <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
   "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
  <html>
    <head>
      <title>"} + obj.status + " " + obj.response + {"</title>
      <style type="text/css">
      #page {width: 400px; padding: 10px; margin: 20px auto; border: 1px solid black; background-color: #FFF;}
      p {margin-left:20px;}
      body {background-color: #DDD; margin: auto;}
      </style>
    </head>
    <body>
    <div id="page">
    <h1>This page is not available</h1>
    <p>Sorry, not available.</p>
    <hr />
    <h4>Debug Info:</h4>
    <pre>Status: "} + obj.status + {"
Response: "} + obj.response + {"
XID: "} + req.xid + {"</pre>
      </div>
    </body>
   </html>
  "};
  return(deliver);
}

Varnish 2.x example

When Varnish has been set-up you need to include Plone virtual hosting rule in its configuration file.

If you want to map Varnish backend directly to Plone-as-a-virtualhost (i.e. Zope's VirtualHostMonster is used to map site name to Plone site instance id) use req.url mutating.

The following maps the Plone site id plonecommunity to the plonecommunity.mobi domain. Plone is a single Zope instance, running on port 9999.

Example:

backend plonecommunity {
        .host = "127.0.0.1";
        .port = "9999";
}

sub vcl_recv {
        if (req.http.host ~ "^(www.)?plonecommunity.mobi(:[0-9]+)?$"
            || req.http.host ~ "^plonecommunity.mfabrik.com(:[0-9]+)?$") {

                set req.backend = plonecommunity
                set req.url = "/VirtualHostBase/http/" req.http.host ":80/plonecommunity/VirtualHostRoot" req.url;
                set req.backend = plonecommunity;
        }
}

Varnishd port and IP address to listen

You give IP address(s) and ports to Varnish to listen to on the varnishd command line using -a switch.

Edit /etc/default/varnish:

DAEMON_OPTS="-a 192.168.1.1:80 \
             -T localhost:6082 \
             -f /etc/varnish/default.vcl \
             -s file,/var/lib/varnish/$INSTANCE/varnish_storage.bin,1G"

Sanitizing cookies

Any cookie set on the server side (session cookie) or on the client-side (e.g. Google Analytics Javascript cookies) is poison for caching the anonymous visitor content.

HTTP caching needs to deal with both HTTP request and response cookie handling

  • HTTP request Cookie header. The browser sending HTTP request with Cookie header confuses Varnish cache look-up. This header can be set by Javascript also, not just by the server. Cookie can be preprocessed in varnish's vcl_recv step.
  • HTTP response Set-Cookie header. This sets a server-side cookie. If your server is setting cookies Varnish does not cache these responses by default. Howerver, this might be desirable behavior if e.g. multi-lingual content is served from one URL with language cookies. Set-Cookie can be post-processed in varnish's vcl_fetch step.

Example of removing all Plone-related cookies, besides ones dealing with the logged in users (content authors):

sub vcl_recv {

  if (req.http.Cookie) {
      # (logged in user, status message - NO session storage or language cookie)
      set req.http.Cookie = ";" req.http.Cookie;
      set req.http.Cookie = regsuball(req.http.Cookie, "; +", ";");
      set req.http.Cookie = regsuball(req.http.Cookie, ";(statusmessages|__ac|_ZopeId|__cp)=", "; \1=");
      set req.http.Cookie = regsuball(req.http.Cookie, ";[^ ][^;]*", "");
      set req.http.Cookie = regsuball(req.http.Cookie, "^[; ]+|[; ]+$", "");

      if (req.http.Cookie == "") {
          remove req.http.Cookie;
      }
  }
  ...

# Let's not remove Set-Cookie header in VCL fetch
sub vcl_fetch {

    # Here we could unset cookies explicitly,
    # but we assume plone.app.caching extension does it jobs
    # and no extra cookies fall through for HTTP responses we'd like to cache
    # (like images)

    if (!beresp.cacheable) {
        return (pass);
    }
    if (beresp.http.Set-Cookie) {
        return (pass);
    }
    set beresp.prefetch =  -30s;
    return (deliver);
}

The snippet for stripping out non-Plone cookies comes from http://www.phase2technology.com/node/1218/

That article notes that "this processing occurs only between Varnish and the backend [...]; the client, typically a user's browser, still has all the cookies. Nothing is happening to the client's original request." While it's true that the browser still has the cookies, they never reach the backend and are therefor ignored.

Another example how to purge Google cookies only and allow other cookies by default:

sub vcl_recv {
    # Remove Google Analytics cookies - will prevent caching of anon content
    # when using GA Javascript. Also you will lose the information of
    # time spend on the site etc..
    if (req.http.cookie) {
       set req.http.Cookie = regsuball(req.http.Cookie, "__utm.=[^;]+(; )?", "");
       if (req.http.cookie ~ "^ *$") {
           remove req.http.cookie;
       }
     }
     ....

Do not cache error pages

You can make sure that Varnish does not accidentally cache error pages. E.g. it would cache front page when the site is down:

sub vcl_fetch {
    if ( beresp.status >= 500 ) {
        set beresp.ttl = 0s;
        set beresp.cacheable = false;
    }
    ...
}

More info

Custom and full cache purges

Below is an example how to create an action to purge the whole Varnish cache.

First you need to allow HTTP PURGE request in default.vcl from localhost. We'll create a special PURGE command which takes URLs to be purged out of the cache in a special header:

acl purge {
    "localhost";
    # XXX: Add your local computer public IP here if you
    # want to test the code against the production server
    # from the development instance
}
...

sub vcl_recv {
    ...
    # Allow PURGE requests clearing everything
    if (req.request == "PURGE") {
        if (!client.ip ~ purge) {
            error 405 "Not allowed.";
        }
        # Purge for the current host using reg-ex from X-Purge-Regex header
        purge("req.http.host == " req.http.host " && req.url ~ " req.http.X-Purge-Regex);
        error 200 "Purged.";
    }
}

Then let's create a Plone view which will make a request from Plone to Varnish (upstream localhost:80) and issue the PURGE command. We do this using the Requests Python library.

Example view code:

import requests

from Products.Five import BrowserView
from requests.models import Request


class Purge(BrowserView):
    """
    Purge upstream cache from all entries.

    This is ideal to hook up for admins e.g. through portal_actions menu.

    You can access it as admin::

        http://site.com/@@purge

    """

    def __call__(self):
        """
        Call the parent cache using Requets Python library and issue PURGE command for all URLs.

        Pipe through the response as is.
        """

        # This is the root URL which will be purged
        # - you might want to have different value here if
        # your site has different URLs for manage and themed versions
        site_url = self.context.portal_url() + "/"

        headers = {
                   # Match all pages
                   "X-Purge-Regex" : ".*"
        }

        resp = requests.request("PURGE", site_url + "*", headers=headers)

        self.request.response["Content-type"] = "text/plain"
        text = []

        text.append("HTTP " + str(resp.status_code))

        # Dump response headers as is to the Plone user,
        # so he/she can diagnose the problem
        for key, value in resp.headers.items():
            text.append(str(key) + ": " + str(value))

        # Add payload message from the server (if any)

        if hasattr(resp, "body"):
            text.append(str(resp.body))

Registering the view in ZCML:

<browser:view
        for="Products.CMFPlone.interfaces.IPloneSiteRoot"
        name="purge"
        class=".views.Purge"
        permission="cmf.ManagePortal"
        />

More info

Round-robin balancing

Varnish can do round-robin load balancing internally. Use this if you want to distribute CPU-intensive load between several ZEO front end client instances, each listening on its own port.

Example:

# Round-robin between two ZEO front end clients

backend app1 {
    .host = "localhost";
    .port = "8080";
}

backend app2 {
    .host = "localhost";
    .port = "8081";
}

director app_director round-robin {
    {
        .backend = app1;
    }
    {
        .backend = app2;
    }
}

sub vcl_recv {

if (req.http.host ~ "(www\.|www2\.)?app\.fi(:[0-9]+)?$") {
    set req.url = "/VirtualHostBase/http/www.app.fi:80/app/app/VirtualHostRoot" req.url;
    set req.backend = app_director;
}