This document is aimed at developers and system administrators who intend to install and configure the WURFL InFuze Module
for Varnish-Cache on Unix, Linux, and other Unix-based systems such as FreeBSD.
In the rest of the documentation, we will refer to the module as the WURFL VMOD
(Varnish Module).
In order for the Module to work it is ESSENTIAL that the libwurfl
library is installed on your system.
libwurfl
is provided in your Customer Vault/FileX.
If you have not already installed libwurfl, instructions can be found here. Release notes for each API can be found here.
The WURFL Varnish module is available for all major Varnish Cache versions.
You may already have an instance of Varnish Cache running on your system. In this case, we recommend that you stop any running Varnish Cache instance and check if you have the correct version of Varnish Cache installed.
__Warning:__ Be aware that some older linux distributions, such as Fedora 15, may install older versions of Varnish library which are NOT compatible with the WURFL C library release. Please make sure that you are running the correct version of Varnish software before installation of the WURFL VMOD.
You can get the Varnish Cache software from packagecloud.io, chosing the desired repo and then
following the installation instructions related to DEB
packages.
Once you have obtained the WURFL Varnish module DEB
package from ScientiaMobile (be sure to get the module related to the Varnish Cache software installed),
you can install it with:
sudo dpkg -r varnish-mod-wurfl
sudo dpkg -i varnish-mod-wurfl-x.y.z.0.varnish-5.2.1.x86_64.deb
Note:
VARNISH MODULE INSTALLATION FOLDER
: Please note that the WURFL Varnish moduleDEB
package will install the module.so
file in/usr/lib/varnish/vmods/
and will try to create symbolic links in/usr/lib/x86_64-linux-gnu/varnish/vmods
and/or/usr/local/lib/varnish/vmods
folders.
If your Varnish installation looks for vmods in different folders than/usr/lib/x86_64-linux-gnu/varnish/vmods
and/usr/local/lib/varnish/vmods
,
on Varnish startup you will see the following error :..../vmods/libvmod_wurfl.so: open shared object file: No such file or directory
.
In this case, please create a symbolic link to/usr/lib/varnish/vmods/libvmod_wurfl.so
in your Varnish vmods folder.
You can get the Varnish Cache software from packagecloud.io, chosing the desired repo and then
following the installation instructions related to RPM
packages.
Once you have obtained the WURFL Varnish module RPM
package from ScientiaMobile (be sure to get the module related to the Varnish Cache software installed),
you can install it with:
sudo rpm -e varnish-mod-wurfl
sudo rpm -i varnish-mod-wurfl-1.9.4.0.varnish-5.2.1.x86_64.rpm
Note:
VARNISH MODULE INSTALLATION FOLDER
: Please note that the WURFL Varnish moduleRPM
package will install the module.so
file in/usr/lib/varnish/vmods/
and will try to create symbolic links in/usr/lib64/varnish/vmods
and/or/usr/local/lib/varnish/vmods
folders.
If your Varnish installation looks for vmods in different folders than/usr/lib64/varnish/vmods
and/usr/local/lib/varnish/vmods
,
on Varnish startup you will see the following error :..../vmods/libvmod_wurfl.so: open shared object file: No such file or directory
.
In this case, please create a symbolic link to/usr/lib/varnish/vmods/libvmod_wurfl.so
in your Varnish vmods folder.
The installation process is now complete. Make sure to check the Configuration Guide
and WURFL Varnish Module Examples
sections to verify that everything was installed correctly.
Warning:
VARNISH TUNING
: Sincelibwurfl 1.8.3.1
the Varnish run time parameterthread_pool_stack
needs to be tuned to meet WURFL requirements.
Its value should be changed to at least 96k executing varnishd daemon with the following command:varnishd -p thread_pool_stack=96k
.
For Varnish versions prior to 4.0, the command isvarnishd -p thread_pool_stack=163840
To perform lookups, you will need a copy of your WURFL data snapshot (also referred to as the wurfl.xml
). While there is one included in the release package, it is intended to be a sample and will not contain all of your licensed capabilities. Your licensed WURFL data snapshot can be accessed by following these directions.
Varnish utilizes Varnish Configuration Language (VCL
), a domain-specific language that can be used to define HTTP-request
handling and media caching policies for the Varnish-Cache HTTP accelerator.
For more information on VCL
, please check the Varnish 5 VCL,
Varnish 4.1 VCL,
Varnish 4 VCL or the
Varnish 3 VCL online documentation as well as other
examples of VCL Usage.
Shown below is an example of a varnish.sample.vcl
configuration file for WURFL setup in Varnish 5.
Please refer to the directives guide which explains each element in detail (Table 1), their parameters,
constraints, and default recommended settings. In order to test the correct installation of the WURFL Varnish module,
you can use the VCL script located in /usr/share/wurfl/varnish.sample.vcl
:
vcl 4.0;
import wurfl;
import std;
backend default {
.host = "127.0.0.1";
.port = "8080";
}
sub vcl_init {
### WURFL root definition. User MUST specify this path in order to make WURFL engine correctly start.
wurfl.set_root("/usr/share/wurfl/wurfl.zip");
### WURFL Updater allows for seamless updates of the WURFL engine with new data downloaded from Scientiamobile.
### Updater configuration must be done after wurfl.set_root
### WURFL file should be either .zip or .xml.gz and match wurfl.set_root file type
### Put your personal updater URL taken from the Scientiamobile customer Vault.
### Valid values for the updater checking frequency (how often the updater checks for any new WURFL data file
### to be downloaded and used by the engine) are DAILY,WEEKLY
### Updater log file (wurfl-updater.log) may be found in "wurfl.set_root" folder. The folder and wurfl.zip file (which must already be present for the updater to work, should be writable by Varnish
#wurfl.updater("https://data.scientiamobile.com/xxxxx/wurfl.zip","DAILY");
#wurfl.updater("https://data.scientiamobile.com/xxxxx/wurfl.zip","WEEKLY");
### WURFL patches definition (as much as needed, patches will be applied in the same order as specified).
#wurfl.add_patch("/path/to/patch1.xml");
#wurfl.add_patch("/path/to/patch2.xml");
### WURFL cache: one of the following
wurfl.set_cache_provider_lru(100000);
#wurfl.set_cache_provider_none();
### WURFL user requested static capabilities (as an example, this is not a complete list)
### If you are upgrading from previous versions, remember that since module version 1.8.1.1
### you don't need to specify WURFL mandatory capabilities anymore.
wurfl.add_requested_capability("is_console");
wurfl.load();
### WURFL Updater startup. Must be done after wurfl.load
### With wurfl.updater_start the updater will execute an asynchronous check DAILY or WEEKLY (depending on wurfl.updater parameters)
### With wurfl.updater_runonce the updater will run only once executing a synchronous check
# wurfl.updater_start();
# wurfl.updater_runonce();
if (wurfl.error()) {
std.syslog(3, wurfl.error());
return (fail);
}
}
sub vcl_miss {
### Print requested static capabilities
std.syslog(0, wurfl.get_capability("is_console"));
### Print useful WURFL setup information
std.syslog(0, wurfl.get_api_version());
std.syslog(0, wurfl.get_wurfl_info());
std.syslog(0, wurfl.get_last_load_time());
### Print some virtual capabilities values
std.syslog(0, wurfl.get_virtual_capability("advertised_browser"));
std.syslog(0, wurfl.get_virtual_capability("advertised_browser_version"));
std.syslog(0, wurfl.get_virtual_capability("advertised_device_os"));
std.syslog(0, wurfl.get_virtual_capability("advertised_device_os_version"));
std.syslog(0, wurfl.get_virtual_capability("is_largescreen"));
std.syslog(0, wurfl.get_virtual_capability("is_app"));
return(fetch);
}
sub vcl_pipe {
return(pipe);
}
sub vcl_pass {
return(pass);
}
Syntax | Description | Availability |
---|---|---|
set_root( string ) | Defines the location (path) of the WURFL data file. | 1.4 |
updater( string, string ) | Allows seamless update of WURFL engine with new data downloaded from Scientiamobile.
A call to set_root must precede it.It takes two parameters: • the data url (taken from your personal Scientiamobile Vault account, choosing between two data file types: .zip or .xml.gz )Take care that set_root file type and updater data url file type match so you may need to change the set_root file type accordingly.• the updater checking frequency (how often the updater checks for any new WURFL data file
to be downloaded and used by the engine) which you can choose between DAILY and WEEKLY .In order to let the Updater perform its activities both the set_root folder and file must be writable by varnish.The wurfl-updater.log file in set_root folder will contains details on Updater activity.
|
1.8.3 |
add_patch( string ) | Adds one or more custom patch files to the WURFL repository. | 1.4 |
set_engine_target_default() or set_engine_target_fast_desktop_browser_match() or set_engine_target_high_performance() or set_engine_target_high_accuracy() |
These configuration options are deprecated and will be removed in a future release. |
1.4 |
set_cache_provider_lru() or set_cache_provider_none( int ) |
In order to increase performance while processing real HTTP traffic, we suggest setting up an LRU cache.
The LRU caching strategy will speed up lookup operations on processed User Agents by keeping them in an LRU map.
By default the cache will be set to 30000 entries which accounts for 7 to 10 MB of additional memory usage.
Specific concerns regarding memory usage apart, users are advised to size their cache generously (100,000 or more) to increase performance.
For more information, please see LRU Cache Mechanism. |
1.4 |
add_requested_capability( string ) | Defines one or more WURFL Static Capabilities to be loaded into memory run-time. | 1.4 |
load() | Loads WURFL engine and makes it available to Varnish. | 1.4 |
updater_start() | Starts the WURFL Updater and executes an asynchronous check DAILY or WEEKLY (depending on updater parameters).A load call must precede it in order to correctly start the Updater.
|
1.8.3 |
updater_runonce() | Runs the WURFL Updater only once, executing a synchronous check. A load call must precede it in order to correctly start the Updater. |
1.8.3 |
error() | Returns the last WURFL call error message or empty if no errors were found. | 1.4 |
const char * get_capability( string ) | Returns the WURFL static capability value for the detected device as a zero terminated ASCII string. If the static capability is missing, this function returns 0. | 1.4 |
int get_capability_as_int( string ) | Returns the WURFL static capability value for the detected device as an integer value if the required static capability exists. If the static capability is missing, or the static capability's value is not a valid integer, this function returns 0. | 1.4.4 |
bool get_capability_as_bool( string ) | Returns the WURFL static capability value for the detected device as a boolean value if the required static capability exists. If the static capability is missing, or the static capability's value is not a valid boolean, this function returns 0. | 1.4.4 |
const char * get_virtual_capability( string ) | Returns the WURFL virtual capability value for the detected device as a zero terminated ASCII string. If the virtual capability is missing, this function returns 0. | 1.5.0 |
bool get_virtual_capability_as_bool( string ) | Returns the WURFL virtual capability value for the detected device as a boolean value if the required virtual capability exists. If the virtual capability is missing, or the virtual capability's value is not a valid boolean, this function returns 0. | 1.5.0 |
int get_virtual_capability_as_int( string ) | Returns the WURFL virtual capability value for the detected device as an integer value if the required virtual capability exists. If the virtual capability is missing, or the virtual capability's value is not a valid integer, this function returns 0. | 1.5.0 |
const char * get_original_useragent() | Returns the original useragent coming with this particular web request. | 1.5.1 |
const char * get_normalized_useragent() | Returns the normalized useragent. | 1.5.1.3 |
const char * get_api_version() | Returns the currently used Libwurfl API version. | 1.5.1 |
const char * get_engine_target() | This function is deprecated and will be removed in a future release. |
1.5.1 |
const char * get_wurfl_info() | Returns a string containing informations on the parsed WURFL data file and its full path. | 1.5.1 |
const char * get_last_load_time() | Returns the UNIX timestamp of the last time WURFL has been loaded successfully. | 1.5.1 |
set_useragent_priority_override_sideloaded_browser_useragent() or set_useragent_priority_use_plain_useragent() |
These configuration options are deprecated and will be removed in a future release. |
1.5.2 |
get_useragent_priority_as_string() | This function is deprecated and will be removed in a future release. |
1.5.2 |
Since virtual capabilities are automatically calculated by WURFL, there are no explicit commands to request them.
There are some functions used to retrieve the virtual capabilities values and also some useful WURFL environment configuration information.
Please take a look at the VCL example and the functions table specified above.
If you try the following examples, you should restart Varnish each time a change is made.
Example 1:
In this example, we declare the vcl_recv
subroutine, which is called when the complete request has been received and parsed. Its purpose is to decide whether an HTTP request should be served. Thus, certain conditions should be satisfied prior to serving the request itself. For example, the req
object represents a single request and its parameters can be accessed via "dotted notation".
In our case, we test if the URL of the request is equal to /
(root). Moreover we check if the browser runs on a device which has a resolution height equal to 960 pixels. If both of these conditions are true then we can use the directive set
to assign a specific home page URL (/wide.html
) to the request parameter URL.
The wurfl
object is accessible after all the required variables have been initialized (see the varnish.sample.vcl
script described before for more info). We can query WURFL for a static capability value simply by calling the function get_capability(cap_name)
.
The subroutine vcl_error
is called every time we hit an error. At this point, the request has been cached by Varnish and as a result we have access to the obj
object's variables in order to check if some error has occured.
For example, we can read the status
variable, which contains the HTTP response status code returned by the server, and act accordingly by showing a message to the user which describes the problem or restarts the transaction to connect to the server.
Please note that, in the subroutine vcl_error
, we return the error object to the client, which in turn can take some further actions.
#################################################################
########## Wurfl Static Capability switch example 1 ##############
##################################################################
sub vcl_recv {
if ( req.url == "/" && wurfl.get_capability("resolution_height") == "800") {
set req.url = "/wide.html";
}
}
sub vcl_error {
if (obj.status == 750) {
set obj.http.Location = "/wideversion/";
set obj.status = 302;
return(deliver);
}
}
Example 2:
This example is similar to the previous one and shows how to throw an error explicitly to the client when both conditions are satisfied. Note that in this case, the request will be discarded by Varnish.
##################################################################
########### Wurfl Static Capability switch example 2 #############
##################################################################
sub vcl_recv {
if ( req.url == "/" && wurfl.get_capability("resolution_height") == "800") {
error 750 "Moved Temporarily";
}
}
sub vcl_error {
if (obj.status == 750) {
set obj.http.Location = "/wideversion/";
set obj.status = 302;
return(deliver);
}
}
Example 3:
In this example we call the hash_data()
function, which will hash the data passed as parameter. Using the +
operator, we can concatenate strings and then pass the result to the hash function. In this case, we consider concatenating the requested URL with the Device ID matched by WURFL.
##################################################################
############## Caching only URL + Device ID example 3 ############
#### Content is cached only if the URL+Device_ID is matching #####
##################################################################
sub vcl_hash {
hash_data(req.url+req.http.host+wurfl.get_device_id());
return (hash);
}
You should restart Varnish every time you make a change to the configuration in order to force a reload of the VCL configuration:
pkill varnishd
varnishd -f /usr/share/wurfl/varnish.sample.vcl
Note: In order to use
service varnish start
you should set properly all the required parameters in/etc/sysconfig/varnish
.
Note: Due to a Varnish limitation, the only working log file is the syslog. If you want to check the error messages and other issues> that may have occurred, you should look into
/var/log/messages
.
© 2024 ScientiaMobile Inc.
All Rights Reserved.
NOTICE: All information contained herein is, and remains the property of ScientiaMobile Incorporated and its suppliers, if any. The intellectual and technical concepts contained herein are proprietary to ScientiaMobile Incorporated and its suppliers and may be covered by U.S. and Foreign Patents, patents in process, and are protected by trade secret or copyright law. Dissemination of this information or reproduction of this material is strictly forbidden unless prior written permission is obtained from ScientiaMobile Incorporated.