Security

At Transloadit, we care a great deal about security. We realize that mistakes can happen, but we are always looking for better ways to fix and prevent those. That is why we really appreciate it when you notify us if you find a security issue.

Responsible Disclosure of a Security Vulnerability

If you think you have discovered an issue with our security measures, please contact us at support@transloadit.com. We will then address your concern as quickly as possible.

We don't hand out bounties/goodies as a reward, but if the issue is valid - i.e. when we think it should be fixed and you are the first to report it - you will be awarded a place in our Hall of Fame. For that reason, please also include in your report how you would want to be credited in the Hall of Fame.

You can supply a site and a name or handle, but we draw a line where things get too cluttered or offensive.

Automated tools can generate a lot of noise for our audit tools, so we ask you not to use those.

Please keep in mind that we do have rate-limiting for login failures and many other events, but that it kicks in quite late. This is because some situations/humans are odd :wink: Yet in case of machines, there would have to be countless guesses for this attack vector to become viable. That is why we allow many attempts before deciding to block someone.

Reports for sites or apps that are not maintained by us, such as Intercom, or imagemagick.transloadit.com, will be relayed to their creators, but do not count towards Transloadit HoF ranking.

Hall of Fame

The following security researchers have identified and responsibly disclosed vulnerabilities to us. To show our gratitude, we are listing them in order of contribution.

Thanks to these people, Transloadit is safer today:

  1. Kirtikumar Anandrao Ramchandani (9)
  2. Rishiraj Sharma — Team Securityescape (9)
  3. Team SecurityescapeEscape (6)
  4. Muhammad Talha Khan (5)
  5. Faissal Ait Hamou (3)
  6. Koutrouss Naddara (3)
  7. Shivankar Madaan (2)
  8. Niharika Sharma (2)
  9. Zee Shan (2)
  10. Daniyal Nasir (2)
  11. Sreehari Haridas (2)
  12. websecresearch.com (2)
  13. Joel Melegrito (2)
  14. Arvind Singh Shekhawat — Team Securityescape (2)
  15. Mukesh Dhama — Team Securityescape (2)
  16. Mazen Gamal Mesbah (2)
  17. Josh Rosen (2)
  18. Shuaib Abidemi Oladigbolu (2)
  19. Vasantha Kumar .S.P. from INFOZIANT
  20. Mohammed Israil (1)
  21. S Naveen Kumar (1)
  22. B.Dhiyaneshwaran (1)
  23. Lawn (1)
  24. Ismail Tasdelen (1)
  25. Sahil Mehra (1)
  26. Nathu Nandwani (1)
  27. Sarang Dabadghao (1)
  28. Prathamesh Joshi (1)
  29. Nikhil Kumar (from Adayptus Consulting) (1)
  30. Monika Talekar (1)
  31. Praful Pradeep Kumar Bhandari (1)
  32. Vijay.S (TwinTech Solutions) (1)
  33. Pradipta Das (1)
  34. Abin Joseph (1)
  35. Joel Melegrito (1)
  36. Himanshu Rahi (1)
  37. Pal Patel (1)
  38. Shivam Kamboj (1)
  39. Rico A. Silvallana (1)
  40. Amjad Kabaha (1)
  41. Shailesh Wagh (1)
  42. Akansha Kesharwani — Payatu Technologies (1)
  43. Md. Nur A Alam Dipu (1)
  44. Kacper Kwapisz (1)
  45. Wai Yan Aung (1)
  46. Florian Courtial (1)
  47. Muhammad Hayat (1)
  48. Pradeep Kumar (1)
  49. Pravin Nagare (1)
  50. Suraj Mulik (1)
  51. Latish Danawale (1)
  52. Kamran Saifullah (1)
  53. Shivam Kumar Agarwal (1)
  54. Fida Hussain (1)
  55. Waqar Vicky (1)
  56. Alec Blance (1)
  57. Ali Wamim Khan (1)
  58. Tayyab Qadir (1)
  59. Jay Jani (1)
  60. Bougioukas Dimitrios (1)
  61. Shawar Khan (1)
  62. Karim Mohamed Ahmed (1)
  63. Lalith (1)
  64. Faisal Ahmed (1)
  65. Nithish M. Varghese (1)
  66. Lawrence Amer (1)
  67. Vishnu Vardhan Reddy (1)
  68. Sane Sindhuja Reddy (1)
  69. SaifAllah benMassaoud (1)
  70. Arun Kumar Agrawalla (1)
  71. Muhammad Osama (1)
  72. Osama Mahmood (1)
  73. Ahmed Jerbi (1)
  74. Pratik Panchal — Infobit Technologies (1)
  75. Ankit Bharathan — Provensec Labs (1)
  76. Ashish Pathak (1)
  77. Provensec labs (1)
  78. Pflash Punk (1)
  79. Aaditya Purani (1)
  80. Jay Patel (1)
  81. Abk Khan — Team MaDLeeTs (1)
  82. Karim Rahal — Karim MTV (1)
  83. João Pina (1)
  84. Muhammad Zeeshan (1)
  85. Ketan Patil (1)
  86. Shahmeer Baloch (1)
  87. Hamid Ashraf (1)
  88. Waseem Ullah Siddiqui
  89. Hammad Qureshi (1)
  90. Marius Kleidl (1)
  91. Yogesh Modi — Team Securityescape (1)
  92. Nagasahas Dasa (1)
  93. Soufiane Ouha (1)
  94. Mohammed Fayez Albanna (1)
  95. Suhas Sunil Gaikwad (1)
  96. Simone Memoli (1)
  97. Rakesh Singh (1)
  98. Uttam Soren (1)
  99. Salman Khan Champion (1)
  100. Daksh Patel (1)
  101. Sumit Shinde (1)
  102. Jerold Camacho (1)
  103. Ruslan Khakimov (1)
  104. Jitendra Jaiswal (1)
  105. Ashutosh Kumar (1)
  106. Ali BawazeEer (1)
  107. Ashish Kunwar (2)
  108. Shwetabh Suman (1)
  109. Piyush kumar (1)
  110. Sreedeep.Ck Alavil (1)
  111. Ifrah Iman (1)
  112. Guhan Raja.L (1)
  113. Akalanka Ekanayake (1)
  114. Vyshnav NK (1)
  115. Steven Hampton (1)
  116. Kaushik Sardar (1)
  117. Ratnadip Gajbhiye (Mr.Ch4rLi3) (1)
  118. Monika Talekar (1)
  119. Abhaychandra Bhaskar Chede (1)
  120. Nikhil Sahoo (1)
  121. Ipsita subhadarshan sahoo (1)

Ordered by number of quality reports and time of reporting (recent first).

Is your name missing from this list or do you want to be credited differently? Let us know.

More Information

If you have questions about Transloadit security in general, here are a few Security-related questions and their answers from our FAQ:

  • Transloadit uses UUIDv4 for generating these IDs randomly. Guessing, or generating a UUID that matches one of ours, would be as probable as generating a collision. This is so improbable that it is not considered not a viable attack vector.

    Since we keep around five million Assemblies in active storage at any given time, the chances are admittedly five million times more likely to generate a collision. That being said, since we rate-limit to 250 operations per minute, it would still take machines longer than mankind has existed on earth to generate enough UUIDs to have a 50% probability that one of those will match a UUID that Transloadit has once generated. We deem this far from being a viable attack vector.

    For files, the window gets even smaller again as we remove them after 24 hours. A few reasons for why we choose to do this are outlined here.

    Beyond the guessing of files or Assembly URLs, it is of course a concern that these addresses would leak somehow. We consider an Assembly ID and file URL private. They are a secret shared between Transloadit, our customer, and depending on your integration, the specific end-user for whom the customer is supplying the files and running the Assembly on your behalf.

    This communication between these parties happens over HTTPS, for which we have A+ grading on SSL Labs across the board. If HTTPS is used for integration with Transloadit and the end-user for any request involved, the URLs to Assemblies and files can not leak beyond these trusted parties, to the likelihood of becoming a viable attack vector.

    Then there is Transloadit to look at as trusted party. Our policy is that only our trusted core-team-members have access to these files for debugging purposes. We receive millions of files every day and they are just UUIDs to us until a customer asks us to take a closer look.

    We run all our processes as non-privileged users, injecting secrets, so if an attacker has possessed these secrets, that means they have gotten root access to our machines somehow. In this case, encryption of the file buckets would not suffice, as with the credentials acquired to get full access to the bucket, the attacker almost certainly also has access to our decryption keys, as is the case when we regard Amazon as hacked. Luckily, both Amazon and Transloadit have a very high focus on keeping our systems secure. But it is true that anyone who would give you a 100% security guarantee here, does in fact, not quite understand security, and you would be wise to steer clear.

  • Our platform is highly volatile in the sense that we'll have 1000 servers online today that will be gone tomorrow. Trying to keep your firewalls up to date with this pace is asking for dropped connections.

    We don't funnel outgoing connections (e.g. /sftp/store or Notifications or /http/import) through one point because of performance and SPOF reasons. Using a fleet of proxies that scale along with load puts us back in the same problem, and using NAT has prohibitive limitations performance-wise. The trade-off of our decision to have maximum reliability & throughput is that our outgoing IPs change rapidly.

    Ingress (Transloadit pinging your server, acquiring results)

    The 'best' we can do is give you your region's Amazon ranges, but obviously you will be whitelisting a lot more than you bargained for. On the other hand you'll still rule out 99% of the internet, so for some less critical use cases it could be viable. We'll list them just in case (updated: 2016-07-22-22-36-08 ):

    • 46.51.128.0/18
    • 46.51.192.0/20
    • 46.137.0.0/17
    • 46.137.128.0/18
    • 52.16.0.0/15
    • 52.18.0.0/15
    • 52.30.0.0/15
    • 52.48.0.0/14
    • 52.95.244.0/24
    • 52.95.255.64/28
    • 52.208.0.0/13
    • 54.72.0.0/15
    • 54.74.0.0/15
    • 54.76.0.0/15
    • 54.78.0.0/16
    • 54.154.0.0/16
    • 54.155.0.0/16
    • 54.170.0.0/15
    • 54.194.0.0/15
    • 54.216.0.0/15
    • 54.220.0.0/16
    • 54.228.0.0/16
    • 54.229.0.0/16
    • 54.246.0.0/16
    • 54.247.0.0/16
    • 79.125.0.0/17
    • 176.34.64.0/18
    • 176.34.128.0/17
    • 185.48.120.0/22

    An up to date Amazon IP list is also available, that page also lists a JSON variant for automation.

    A better solution you could implement, if security is paramount, is to setup a server outside of your trusted zone that we'll push updates to. Machines inside your trusted zone could then pull updates from this machine using rsync, a database, or ZeroMQ (collect Notifications and have your trusted zone eat through this queue). This way you will only have to deal with a limited set of known IPs. Additionally, this will cover the security risk that, should our machines ever get compromised, there is a whitelisted connection straight into your trusted zone. We'd push file results to an S3 bucket with append-only access, and you can then from inside your DMZ safely pull them in.

    For a near-real-time approach, something like HAProxy (directly forward traffic into your trusted zone), direct routing, or SSH tunnelling the notify_url into your DMS will work.

    Egress (Client-side integrations talking to Transloadit)

    If you're deploying a client-side integration of Transloadit (like Uppy) with a corporation that puts restrictions on internet use, please whitelist these domains:

    • *.transloadit.com
    • *.*.transloadit.com
    • s3.amazonaws.com
    • s3-eu-west-1.amazonaws.com
  • If you want us to store files in your S3 bucket, it is recommended to save the credentials in a Template in your account. We keep this Template encrypted in our database.

    The keys needed to decrypt these are injected into process memory by another system user. This means that if someone was able to exploit the user under which we run our API or website processes, they could still not access the keys. If they tried to change our code to display or send the credentials, they would have to restart the service (not permitted under that user) and access the key files (also not permitted).

    If our servers are rooted, it is a different story. This is why we use firewalls, use protected SSH keys, and limit our sudo, but as any expert will tell you, 100% security is a myth and it is better to prepare for the worst.

    This is why we recommend to create IAM policies that only have Put and List permissions on your buckets (for an up to date and precise list of the required permissions, please check the S3 Store documentation), and let Transloadit use that for writing only. So if your credentials were ever stolen and the criminals managed to decrypt them as well - they would still only be able to add more files to this particular bucket, until we notice and intervene.

    While, as said, 100% security is a myth, our security philosophy is to make it as hard as possible to for anyone to gain access to your credentials and the keys necessary to decrypt them, and if they do manage to acquire them, to make them as useless as possible.

  • First of all, for signature authentication and all other things that need to be secure we do not use MD5. We provide MD5 hashes for files as a means to detect duplicates from trusted sources, as well as for other purposes where there’s no rewarding use case for the attack vector of calculating colliding hashes.

    Since MD5 is both faster to calculate and even more widely available than, for instance, SHA1, a deliberate trade-off was made by providing MD5 hashes for encoding results. It goes without saying that, since MD5 is not secure, you should not be using these hashes as building blocks for anything that is security sensitive in your application.

Blog posts about security

Over the years we wrote the following posts about security on our blog:

And here are the Security related pages in our documentation: