Persisting the Storage Folder

The storage folder by default, holds several “generated” essentials of your Laravel application.

It’s the default burrow for session, cache, and file data amongst others. If you’d opt to persist data on this folder, you’ll have to mount a volume to it.

The Steps

Make sure you are in your Laravel Fly App’s directory
```
cd <laravel-fly-configured-app>
```

Then, check the number of machines currently available for your Fly App:

fly status

In this example today, the app is deployed in the AMS region, and has the default number of two machines created for it:


Machines
PROCESS ID              VERSION REGION  STATE   ROLE    CHECKS  LAST UPDATED         
app     5683945bd46448  1       ams     started                 2023-10-03T09:54:04Z
app     e82d922f010908  1       ams     started                 2023-10-03T09:54:25Z

To mount a volume named "storage_vol" to this Fly App, you’ll have to create two "storage_vol" volumes in the AMS region, one for each machine:
```
fly volumes create storage_vol --region ams --count 2 
```
Running the command above should create two separate volumes with the name “storage_vol”, in the AMS region for your Laravel Fly App.
Next, revise your Laravel Fly App’s fly.toml file to mount the Volumes above to each machine’s storage folder:
```
[mounts]
  source="storage_vol"
  destination="/var/www/html/storage"
```
Mounting a Volume to a folder will initially erase any item the folder contains during the first time the Volume is mounted for the folder.

For example, Laravel’s storage folder contains subfolders: app, framework, and logs. Mounting a volume to the storage folder erases these directories, and leaves behind a sole item paradoxically named as “lost+found”.

But, you wouldn’t want to only be left with “lost+found” in your storage folder. You’d want to still have the necessary files and directories in there for successful session, views, caching, and file storage compliance with Laravel’s default configuration.
To fix the little storage-content-erasure issue as stated in the callout above, please go ahead and make a copy of your storage folder in a “backup” folder. You can name this directory “storage_”.
```
cp -r storage storage_
```
You’ll later use this folder to copy over its contents to the volumized storage folder.
Next create a Startup Script that will initialize the volumized storage folder’s contents.
```
touch .fly/scripts/1_storage_init.sh
```
Start up scripts are run in numeric-alphabetical order. Naming 1\_storage\_init.sh makes sure it is the first script run. Otherwise, naming the file as storage_init.sh alone would’ve moved the caches.sh script above it, and would’ve executed before storage initialization happened. One of the commands in the caches.sh will not have worked properly, due to a lack of properly initialized storage directory.

On to the content of the Start Up script:
```
FOLDER=/var/www/html/storage/app
if [ ! -d "$FOLDER" ]; then
    echo "$FOLDER is not a directory, copying storage_ content to storage"
    cp -r /var/www/html/storage_/. /var/www/html/storage
    echo "deleting storage_..."
    rm -rf /var/www/html/storage_
fi
```
So what happened above?
- The condition statement checks if the app folder does not exist in the volumized storage folder. If it does not exist, it copies over the contents of the storage_ folder to the volumized storage folder.
Finally, deploy your Laravel Fly App!
```
fly deploy
```

Fly Volumes do not automatically sync their data with each other. Please remember to create the appropriate data-replication logic if your Fly App will be using more than one volume instance, and if your app requires data available across its Volumes.

Possible Errors

Error not enough volumes named `<volume_name>` (1) to run `(<n>)` processes

The above error can come up after configuring your volume in fly.toml and executing fly deploy.

It can mean that there are <n> processes configured in your fly.toml trying to use the volume! Take note however, that a Volume can only be used by one at any given time.

One way to fix this issue is to separate each process into different Fly.io apps. Of course, separate application per process might require inter-communication between the applications.

Fly.io applications can easily communicate with each other over a private network. Not only that, but Fly.io also offers the fly-replay response header which can be used to “redirect” request from one application to another, and return response from the correct application.

Edit this page on GitHub