Following this post and this, here's my situation:
Users upload images to my backend, setup like so: LB -> Nginx Ingress Controller -> Django (Uwsgi). The image eventually will be uploaded to Object Storage. Therefore, Django will temporarily write the image to the disk, then delegate the upload task to a async service (DjangoQ), since the upload to Object Storage can be time consuming. Here's the catch: since my Django replicas and DjangoQ replicas are all separate pods, the file is not available in the DjangoQ pod. Like usual, the task queue is managed by a redis broker and any random DjangoQ pod may consume that task.
I need a way to share the disk file created by Django with DjangoQ.
The above mentioned posts basically mention two solutions:
-solution 1: NFS to mount the disk on all pods. It kind of seems like an overkill since the shared volume only stores the file for a few seconds until upload to Object Storage is completed.
-solution 2: the Django service should make the file available via an API, which DjangoQ would use to access the file from another pod. This seems nice but I have no idea how to proceed... should I create a second Django/uwsgi app as a side container which would listen to another port and send an HTTPResponse with the file? Can the file be streamed?
Third option: don't move the file data through your app at all. Have the user upload it directly to object storage. This usually means making an API which returns a pre-signed upload URL that's valid for a few minutes, user uploads the file, then makes another call to let you know the upload is finished. Then your async task can download it and do whatever.
Otherwise you have the two options correctly. For option 2, and internal Minio server is pretty common since again, Django is very slow for serving large file blobs.
I ran into the problem where Heroku doesn't update my GitHub repository (or say static filesystem) when a blog post (including pictures) is created from the website.
Other images survive, whilst the ones saved in my filesystem with the server running on heroku, disapper.
I found this on their documentation.
The Heroku filesystem is ephemeral - that means that any changes to the filesystem whilst the dyno is running only last until that dyno is shut down or restarted.
I'm still confused why not all the pictures disappear and only those added later do.
Is AWS S3 a solution for this? If it is, how can I represent my filesystem using buckets?
Say, for the Blog Post 1 I have 2 picture resolutions, which means storing the files in different folders corresponding to those resolutions.
---1920x1920
-----picture.jpg
---800x800
-----picture.jpg
Does that mean I have to create 2 buckets named 1920x1920 and 800x800 or is there a better way of handling them?
Is AWS S3 a solution for this?
S3 is the recommended solution for this, and the configuration is documented in Heroku DevCentre with specfic instructions for uploading from Python.
Note these Python instructions use the Direct Upload approch: Have the flask app generate a pre-signed URL, which is then passed back to the client Javascript code, so that the user's browser can make the upload to S3 directly. The resulting S3 URL of the image, is then put into a hidden element in the form, which is then received by your app on form submit.
The fact that you have separate image sizes suggests your app does some processing (maybe with PIL) to get these thumbnails. In which case it may be easier to use the Pass-Through approach where your app implements its own upload mechanism, does the processing and then uploads the thumbnails to S3 (The upload to S3 part is well document, such as in this SO thread).
The Pass-Through method carries the warning that this may cause blocking of a single threaded worker. If your site gets a volume of requests that causes this to be an issue, you may need to increase the number of gunicorn workers, or change to a worker type that supports concurrency (This github post has some useful commands/info on concurrent worker types).
The best way to implement this whole thing (although the requirement for a redisgo dyno and worker dyno may push you into the paid teir) may be with Background Tasks using rq. You use the Direct-Upload approach above to upload the original image, then have a background job download that, do the resizing, and put the resulting thumbnails back onto S3.
Does that mean I have to create 2 buckets named 1920x1920 and 800x800 or is there a better way of handling them?
Have one Bucket for the entire app, and just include forward slashes in the object's key to mimic a subdirectory structure.
I am encountering what I believe to be permission issues when trying to deploy a Drupal application onto Elastic Beanstalk
I followed this tutorial to get Drupal up and running: http://comm-press.de/en/blog/drupal-climbs-aws-elastic-beanstalk
I am using a Postgres database and I am entering the correct credentials when filling out the forms on install.php, without error.
Any subsequent deploys after the initial deploy brings me back to install.php. After entering in my database information, I get this message, telling me Drupal is already installed (which it is).
http://i.imgur.com/N6KDvvo.png
Why does my site get redirected to install.php after 'eb deploy'? What permissions should I set my drupal folder such that settings.php and /sites/default/files is generated?
The install state is controlled by the DB-- if Drupal bootstraps with no DB information, you are routed to the code that asks you for it.
I was able to bypass this part by setting up a AWS RDS DB and connecting all instances to it.
--But, wait, there's more. Now that having all the instances reading from the same DB has squashed most of the concurrency problems between instances*. Go ahead, try and add a photo to your admin profile. I will wait. Yep, most of the time you'll get the wrong instance and the single image on one instance is not shown on all instances.
I am working on solving that problem with a start up & cron job script that loads updates to resources from the AWS S3 service.
Step A load code into S3
Step B set an accessible timestamp for $lastModified to now()
Step 1 wget/curl a timestamp of the last remote modification ($lastModified)
Step 2 compare the local last updated stamp ($lastUpdated) to remote last modified timestamp
Step 3 if ($lastModified == $lastUpdated) {die} else {update incremental changes && set $lastUpdated = $lastModified}
Watch that first incremental update, it's a doozy.
So... additionally I should mention that we are installing completely vanilla drupal when we instantiate an image, as part of the Docker file from drupal the drupal apache image the last thing the Docker file runs is a setup script.
Elastic Beanstalk sets Environment variables-- some of those variables are the amazon access key id and access secret key.
I curl an IP only available inside of Elastic Beanstalk curl -v 169.254.169.254
From that output in the setup script I can tell if I am local AWS EB or in AWS EB. That allows me to conditionally change certain configurations-- like connecting RDS or a local MySQL DB to Drupal.
The setup script uses aws cli to pull from S3 (sync) to add, replace, update everything in the webroot turning the instance into a setup Drupal installation as far as file level assets go.
sed and service reload is done a lot. Elasticache vs local Redis...
last we start the web server in the foreground && tail -f /dev/null so the container doesn't immediately close.
Drupal is just for static assets pages and a header/menu/footer wrapper for our web app (templates are served..., Twig/JS fills the template in with data). Authentication happens via API-- not even using 90% or so of the goodness in Drupal...
Incremental changes are pulled by comparing hash values and acting to run the update process if they are different.
Suppose I create a site in Wordpress, which is running on Elastic Beanstalk. Now, on the running app I will create posts /pages, upload images, etc. That is, some data, videos, files and records in a database will be added to the running application.
3 questions:
If WordPress is running on Elastic Beanstalk with multiple Amazon EC2 instances actually running my WordPress install, then will those files propagate automatically to all running instances? And will this also happen, if a new EC2 instance is fired up - for example, to handle increased load?
From what I see in AWS console, I can deploy different versions of an app-- but as per scenario above, if I deploy a new version, wont I lose all the files uploaded directly into running app (i.e. files and database records)? How do I keep those and at the same time deploy a new version of my app?
The WordPress team keeps issuing upgrades. Can I directly upgrade my running WordPress install, through the web interface? Or do I have to first upgrade my local version of WordPress, and then upload the new version of the app to Beanstalk? If I have to upgrade my local version and then upload, then again I am back to point 1, i.e. changes made by users directly to the older version of running app. How do I preserve those changes?
I've been working on this as well, and have learned a couple of things that are relevant here -- your question about uploads in particular has been on my mind:
(1) The best way to handle uploads, it seems to me, is to either go the NFS/NAS route like you suggest, but one better than that is to use an Amazon S3 plugin for WordPress, so that any uploads automatically copy up to S3 and the URLs in your WordPress media library reflect the FQDN of your bucket and not your specific site. That way you could have one or ten WP nodes in your Beanstalk and media/images are independent of any one of those servers.
(2) You should absolutely be using RDS here. Few things are easier to work with and as stress-free as a Multi-AZ, reserved MySQL RDS instance. Either that or your own EC2 running MySQL that is independent of the Beanstalk, but why run that when RDS is so much easier?
(3) Yes you definitely have to commit changes to your Git repository or local file first (new plugins, changes to themes, WP upgrades) and then upload/install as a revision to the Beanstalk code. Otherwise, all the changes you make via the web interface to one node will never be in the new load for a new node -- in fact you'll have an upgraded database but an older set of code in the Beanstalk application, so it's likely to create errors of some kind or another.
I took an AWS architecture course, and their advice for EC2 and the Beanstalk is to start to think about server instances as very disposable -- so you should try to think about easy ways for your boxes to provision themselves in the bootstrapping process and to take over work for one another without any precious resources on just one box. So losing an instance should never be a big deal. (This is definitely not how we thought in the world of physical servers, where we got everything tweaked 'just so'.)
Good luck!
Well, I'm no expert, but since no one has answered, I'll give it my best shot.
You are absolutely right--kind of. While each EC2 instance does have some local storage, it is destroyed and reset with each new instance. Because of this, Amazon has things like Elastic Block Storage and S3 for persistent files. I don't know how one would configure WP to utilize this, but that will likely be the solution.
I think this problem is solved by my answer to #1. As for the database, all of your EC2 instances should be pulling from the same RDS location. Again, while you could have MySQL running on each EC2 instance, in the interest of persistence, having a separate database makes more sense.
You, again, have most everything right. Local development should always precede live deployment. Upgrading locally then pushing to the live servers will make sure all of your instances remain identical.
Truth-be-told, I am still in the process of learning all of this too, as I said, I'm not an expert. Hopefully someone else will come along and give a more informed answer. However, the key conceptual hurdle here is the idea of elastic scalability--and the salient point of this idea is the separation of elements between what is elastic/scalable/disposable and what is persistent.
Hopefully that helps.
I have deployed a small Wordpress site on EB, S3 and RDS. S3 holds all static data, such as media uploads. This works through a plugin. RDS holds the database. EB holds the latest deployed application. The application is deployed from my dev environment, with a build script. This way, I just have to press one button and I redeploy.
I wrote an article about it here: http://www.cortexcode.com/wordpress-to-aws-code-example/
While it was at first annoying to work with, the speed of AWS is nice and now it's easier than ever. It used to be that I had to upload a bunch of files over FTP, this is way more efficient. :-)
As an addition to all the great answers already:
1) I can highly recommend EFS but also S3 for media files, so they are served from high availability regions in combination with cloudfront. For Wordpress there is one plugin that really speeds up this ( not affiliated to them, just really like the plugin ). There is also an assets plugin, if you'd like to serve JS, CSS files from S3. For the EFS solution, take a look at the AWSlabs docs on git, and specifically this file on how they mount the uploads file.
In general, EBS is really great for Wordpress, but you'll need to think in a different mindset as compared to other hosting solutions ( shared hosting, managed hosting ).
OK I researched a lot on this particular issue, and this is what I learned--
(1) If a wordpress user uploads some files, then his files will be uploaded only to the virtual machine that is actually serving his request at that time. Eg if currently the wordpress site is cloud-deployed and is using 5 virtual machines, now when user makes request he is directed to one virtual machine-- the one with the lowest load at that point... His uploads are stored only on that server. Current Platform-as-a-service solutions (like Amazon Elastic Beanstalk and App Fog) do not have the ability to propagate the changes to all the running instances. Either that (propagate changes to all servers) or use a common storage by all running instances-- these are the only 2 solutions to this problem. (Eg of common storage would be all 5 running virtual machines using Network-Attached-Storage (NAS)... )
(2) With ref to platforms available currently like Amazon Elastic Beanstalk and App Fog, for example, even if user made changes directly to running app- these platforms rely on the local version of code (which the admin deployed initially to cloud)- and there is no way to update the local version of code (on admin's PC) with the changes made by a user to running app-- hence these changes viz, files are lost-- Similarly, changes in database by user to running app are also lost-- unless the admin is using exactly the same database for his local app (that he deployed to cloud)
(3) Any changes to running apps first have to be made to the local app on admin's PC and then pushed to cloud.
I am working on a Cloud PaaS that addresses all these concerns-- viz updates can be made to running apps, code changes made to running app are also updated in code repository accessible by user...The Proof of concept is ready, hopefully it will be as good as I hope it should be :) -- currently the only thing that is actually there is the website (anyacloudpanel.com) -- design work is going on :)
If there is some rule that I should not mention my website( Anya Cloud Panel) -- then I am sorry -- pls feel free to edit and remove my website URL from my answer :)
Thanks,
Arvind.
Deploying WordPress to AWS Elastic Beanstalk does require some change to the normal WordPress deployment as mentioned here a few times. To answer your questions, here is a great tutorial explaining stateless applications and how to deploy to Elastic Beanstalk:
Deploying WordPress to Amazon Web Services AWS EC2 and RDS via ElasticBeanstalk
Be careful if you use a theme from themeforest for example. Some of them are incompatible with wordpress S3 plugin. Then you're screwed, you can not deploy your wordpress on the cloud.
I recently deployed a couple of web applications built using django (on webfaction).
These would be some of the first projects of this scale that i am working on, so I wanted to know what an effective backup strategy was for maintaining backups both on webfaction and an alternate location.
EDIT:
What i want to backup?
Database and user uploaded media. (my code is managed via git)
I'm not sure there is a one size fits all answer especially since you haven't said what you intend to backup. My usual MO:
Source code: use source control such as svn or git. This means that you will usually have: dev, deploy and repository backups for code (specially in a drsc).
Database: this also depends on usage, but usually:
Have a dump_database.py management command that will introspect settings and for each db will output the correct db dump command (taking into consideration the db type and also the database name).
Have a cron job on another server that connects through ssh to the application server, executes the dump db management command, tars the sql file with the db name + timestamp as the file name and uploads it to another server (amazon's s3 in my case).
Media file: e.g. user uploads. Keep a cron job on another server that can ssh into the application server and calls rsync to another server.
The thing to keep in mind though, it what is the intended purpose of the backup.
If it's accidental (be it disk failure, bug or sql injection) data loss or simply restoring, you can keep those cron jobs on the same server.
If you also want to be safe in case the server is compromised, you cannot keep the remote backup credentials (sshkeys, amazon secret etc) on the application server! Or else an attacker will gain access to the backup server.