Installing and configuring Redis for Symfony takes quite a few steps. So many I’d never remember them all. This article is for myself at a later date as well as anyone else who finds it useful. I’ll be updating this article as I learn more.
This article covers installing and configuring Redis for use for both Session storage and Application cache.
If you are using docker once you have started a Redis instance test it by trying to use the cli like so
You should see something similar to this.
This means redis is running on 127.0.0.1 (localhost) on port 6379 which is the default port.
With redis-cli running you can further test with the following.
//set a key and value
set someKey "some value for the key"
//get the value for the key
//view a list of all keys in redis storage
In production you need to install Redis or have access to a server running Redis, I’ll cover that in another article.
Install phpredis extension
You will need to install phpredis php extension and configure it. Before you can even do that though, you will need to install another php module php-dev I am using php 7.4 and Ubuntu so to install that I do this.
apt-get install php7.4-dev
This is needed because phpredis use phpize and phpize is included in php-dev.
If you are using another version of php you can search apt repository for this package like this:
apt search php-dev
or for version specific like this apt search phpver-dev apt search php7.4-dev
This is just the extension for the client to interact with your Redis server wherever it is, either local or remote.
Now you must configure PHP to use this extension.
You could add the needed config values to the php.ini config, but the problem is there are two. Yeah one for the cli and one for fpm. I have an easier solution. Create one file and symlink for both cli and fpm.
You will need both configured. As I found out if you configure only fpm your app will work, but when you go to composer install/update/require etc. you will get a cli error about missing such and blah Redis extension blah blah.
If you are running PHP 7.4 on Linux you will want to create a file in the following /etc/php/7.4/mods-available directory named phpredis.ini with the following
session.save_handler = redis
session.save_path = "tcp://localhost:6379?timeout=3&read_timeout=3"
The way this works is php after it reads the php.ini reads in all of the configuration files ( those with .ini extension) from the conf.d directory for either cli if you are using the command line or from fpm for your app. This makes configuring anything you need for php easier than having to open the giant php.ini file, plus you don’t have to worry about ruining one, which I have done easily. Here is a link to the php docs on configuring and .ini files
Now you must restart php fpm for your app to work. On Ubuntu you can do this.
service php7.4-fpm restart
Configure Symfony for Redis Sessions
Now you must configure some things in Symfony. Part of the following can be found in the docs about caching in a Redis Database here.
From the docs you can see you need set these values inside services.yaml which is in the config directory of your app.
# you can also use \RedisArray, \RedisCluster or \Predis\Client classes
# uncomment the following if your Redis server requires a password
# - auth:
# - '%env(REDIS_PASSWORD)%'
The values for REDIS_HOST, REDIS_PORT and REDIS_PASSWORD should be defined in environmental variables on your system or in .env or in the secrets vault. For testing .env.test.local works.
Now there is still a little more configuring as the docs show in the link above. You need to configure the framework to use Redis for session storage. Open framework.yaml located in config/packages/ and change the handler_id and comment out the save_path file location info like so.
You also need to configure the cache now in cache.yaml if you want to use Redis as a cache for your app. You could configure everything in framework.yaml but it becomes a mess if you do that. Symfony reads all the files recursively located in the config directory, just make sure your yaml structure is correct.
If you open /yourapp/config/packages/cache.yaml you should see something similar already there.
# Unique name of your app: used to compute stable namespaces for cache keys.
# The "app" cache stores to the filesystem by default.
# The data in this cache should persist between deploys.
# Other options include:
# APCu (not recommended with heavy random-write workloads as memory fragmentation can cause perf issues)
# Namespaced pools use the above "app" backend by default
Un-comment the lines shown under Redis section. You will notice a special syntax I am using. I kept messing around until it worked. You might not need to configure the default_redis_provider I need to do more research on that because it seems like that should be covered from the configs above, seems redundant.
Often you may need to know whether a user is logged in or not inside a template to show or not show something. For example you might want to show links to login or register if a user is not logged in but show a link to logout if the user is logged in.
To do this you use is_granted() within a template with one of the following.
Using ROLE_SUPER_ADMIN_1 which is something I made up for my own app to check what type of admin the user is. I don’t really like the IS_AUTHENTICATED_* methods, read more about them in the link below if you want.
Sometimes we make mistakes. When first learning docker we probably make many mistakes and end up with tons of unused docker images.
If you are on Linux like me, you won’t have a desktop dashboard like Mac and Windows get, so things are harder. To see a list of what images you have created you use the following command
docker ps -a
That command will output something like this.
It will show the CONTAINER ID, IMAGE, COMMAND etc. as you can see. To delete an image you use the container id with docker rm like this.
docker rm b5f8fae52bce
I’ve seen older internet posts using the IMAGE value but I had no success with that method. I am guessing something changed. I didn’t even see an example of this I just tried it. I don’t see any mention of this in the docs either. But it works. This part in the getting started intro actually explains it. I think something did change.
First what I wanted to do was create a cookie in a Controller and display a template at the same time. Sort of like when a user visits a page you set a page count or something. The documentation doesn’t really show an example, you are expected to know it via “common sense” apparently according to one smartass.
There is more than one way I have discovered over time. Apparently you can use render the same way I show using renderView.
Below is the Symfony Cookie class create method comment/documentation. This is all of the values you can supply when creating a cookie.
* @param string $name The name of the cookie
* @param string|null $value The value of the cookie
* @param int|string|\DateTimeInterface $expire The time the cookie expires
* @param string $path The path on the server in which the cookie will be available on
* @param string|null $domain The domain that the cookie is available to
* @param bool|null $secure Whether the client should send back the cookie only over HTTPS or null to auto-enable this when the request is already using HTTPS
* @param bool $httpOnly Whether the cookie will be made accessible only through the HTTP protocol
* @param bool $raw Whether the cookie value should be sent with no url encoding
* @param string|null $sameSite Whether the cookie will be available for cross-site requests
* @throws \InvalidArgumentException
Then the cookie will only live/exist until the user closes their browser(unless your browser restores from your last session). You must supply an expires time to make it persist beyond closing the browser. Providing an expires time gives you better control over when the cookie expires due to the above mentioned browser restore issue which will restore cookies that should have died on browser close.
You can also create the cookie then pass it to setCookie() like this.
Here I set the expires to a number, time() returns a linux/unix timestamp and I added 36000 seconds or 10 hours to it. This cookie will exist until the user refreshes their page or clicks a link in 10 hours from creation. However long you want it to live you add that many seconds. Or you could create a date using PHP DateTime as you can pass a DateTime object to the expires position. You then use the methods of DateTime to increase the time to a period in the future and pass the DateTime object after calling the methods to do so.
Side Note : in the above code, you can create a cookie without the $response->setContent() call. I do that with the body tag so that the profiler will show up at the bottom of the page for debugging.
That code goes inside a controller method for the requested route by the way. Usually you use the render() method inside a controller to send a response, which renders the template and sends it in a response. You can also use renderView to do the same thing and capture the value in a variable then use setContent or just make the renderView call right in setContent. I know that works. You can also store the returned value from render the same way. But no matter how you do it, you must return the response object, the very last line. You can find all the methods of the Response class here in the source code.
If you wanted to render a view which requires variables to be sent you do it like this and capture the output of renderView().
Note : do not just use php setcookie or setrawcookie. The reason is they start sending output headers to the browser, which may interfere with how symfony works. You probably won’t notice in a browser, but you may get errors when testing your controllers with functional tests etc.
Personally I created a huge class which extends DateTime which has all kinds of methods for adding days, hours, removing them and doing other math. I’d share it on github but it has bugs since I wrote it way back in version 5 of php in 2012. Some changes were made to DateTime and I haven’t had time to review them all and hunt down the changes that need to be made yet. I’ll probably do it and add it to github eventually. But for now I use time() + seconds. It’s not the best solution but it works and I only need this one cookie.
Most documentation sucks. It seems most authors have this idiotic idea that whatever is in their head, all their knowledge etc. magically transfers to the user as they read it. These are the same kind of idiots you meet in society that drone on about shit they understand to the totally uninitiated like they are talking to a pro. These are the kind of idiots that while talking don’t even have the ability to notice their true live audiences eyes glazing over and heads shaking in confusion. Nope these idiots are comfortable being idiots and we have to suffer. I don’t get it.
Ever been reading along documentation and the author just keeps throwing out concepts/terms left and right without even giving a link, leaving you in the dark like WTF? Isn’t it fun?
Here is a tip to anyone who writes documentation. No matter how stupid simple a concept seems to you, there will be people, many people who have never come across the concept. Instead of being a snobby dickhead with an attitude like “Well you have to have a minimum idea of how blah blah blah works because I do”. You should provide information, links etc. to any concept that anyone may not have encountered. It is not that damn hard and doesn’t take much time and it makes you look much smarter and much less conceited and narcissistic.
This is the exact reason I started this website. So much of the documentation for everything sucks ass. Most documentation I come across these days leaves me saying WTF. Look at openssl or openssh or example so many WTF’s in those docs.
And what is worse is the assholes who write blogs who just copy other blog posts. So you get entire search pages full of shit articles all repeating the same wrong shit that doesn’t work. Then that shit information ends up on stackxxxx.com sites from idiot kids who copy and paste shit for points and don’t care if the answer is right, they just want popular points. I often just avoid stack sites these days. Those sites are only handy if you are doing the most basic simple shit. Once you go into complex issues those sites are wrong answer land 100% of the time.