惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Google DeepMind News
Google DeepMind News
Stack Overflow Blog
Stack Overflow Blog
Hugging Face - Blog
Hugging Face - Blog
博客园_首页
T
The Blog of Author Tim Ferriss
博客园 - 叶小钗
N
Netflix TechBlog - Medium
腾讯CDC
C
Check Point Blog
P
Proofpoint News Feed
Engineering at Meta
Engineering at Meta
GbyAI
GbyAI
S
SegmentFault 最新的问题
F
Fortinet All Blogs
美团技术团队
U
Unit 42
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
博客园 - 司徒正美
F
Full Disclosure
Recorded Future
Recorded Future
D
DataBreaches.Net
博客园 - 【当耐特】
Martin Fowler
Martin Fowler
J
Java Code Geeks
I
InfoQ
Y
Y Combinator Blog
A
About on SuperTechFans
AI
AI
爱范儿
爱范儿
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
Forbes - Security
Forbes - Security
W
WeLiveSecurity
M
MIT News - Artificial intelligence
雷峰网
雷峰网
cs.CV updates on arXiv.org
cs.CV updates on arXiv.org
Simon Willison's Weblog
Simon Willison's Weblog
Schneier on Security
Schneier on Security
The GitHub Blog
The GitHub Blog
Security Archives - TechRepublic
Security Archives - TechRepublic
aimingoo的专栏
aimingoo的专栏
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
G
GRAHAM CLULEY
Know Your Adversary
Know Your Adversary
Latest news
Latest news
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
D
Docker
Recent Commits to openclaw:main
Recent Commits to openclaw:main
量子位
V2EX - 技术
V2EX - 技术
Project Zero
Project Zero

IT Notes - social

IT Notes IT Notes IT Notes IT Notes IT Notes IT Notes IT Notes IT Notes IT Notes IT Notes
IT Notes
Stefano Marinelli · 2025-02-08 · via IT Notes - social

One of the useful yet resource-intensive features of platforms like Mastodon is that they reprocess (and store locally) all multimedia files from other instances.

This behavior is implemented for three valid reasons:

  • To ensure files do not contain malicious code by reprocessing them locally.
  • To prevent all users from all instances from overloading the original instance hosting the media by requesting it repeatedly.
  • To hide the individual IP addresses of users from the original instance.

While this approach has benefits, it also requires significant disk space - often many gigabytes per day.

snac, on the other hand, does not locally process or cache media by default. Instead, media URLs remain unchanged, meaning users will fetch content directly from its original source. This behavior is perfectly fine for many setups (such as mobile devices), but it may not be suitable for everyone.

I suggested a small improvement to snac’s developer, who immediately recognized the benefit for users and implemented an interesting new feature: media proxying via the instance itself.

Enabling Media Proxying in snac

To enable this feature, simply add the following line to your server.json configuration file:

"proxy_media": true

Once enabled, snac will rewrite all media URLs to pass through its own instance. This ensures that original instances will no longer see the IP addresses of individual users, as they will only see the IP of the snac instance itself.

However, this also increases the load and bandwidth consumption of the instance since it must download and forward media every time a user accesses it.

Caching Proxied Media with nginx

To optimize performance, we can configure nginx to cache these proxied files transparently. This way, if multiple users from the snac instance (or the same user at different times) request the same media file, nginx will serve it from the local cache instead of fetching it again. This setup has two key advantages:

  • Reduces traffic to the original instances.
  • Lowers the load on the snac instance, as it won’t need to download the same file repeatedly.

To achieve this, create a cache storage area in nginx by adding the following to the http section:

proxy_cache_path /var/cache/nginx/snac_media levels=1:2 keys_zone=snac_media:10m max_size=1g 
                 inactive=1d use_temp_path=off;

This defines a cache directory at /var/cache/nginx/snac_media, with 10 MB allocated for metadata and a maximum cache size of 1 GB. Cached content will be invalidated and removed after one day, ensuring frequently accessed content (like profile avatars) remains cached while rarely used files get replaced over time.

Next, add the following rule to your snac instance's virtual host configuration:

location ~ ^/.+/(x|y)/ {
    proxy_cache snac_media;
    proxy_pass http://snac-ip:8001;
    proxy_set_header Host $host;
    proxy_cache_valid 200 1d;
    proxy_cache_valid 404 1h;
    proxy_ignore_headers "Cache-Control" "Expires";
    proxy_cache_use_stale error timeout updating http_500 http_502 http_503 http_504;
    proxy_cache_lock on;
    add_header X-Proxy-Cache $upstream_cache_status;
}

After reloading nginx, the cache will start populating as users request media files. Unrequested content will never enter the cache, ensuring efficient storage usage without unnecessary clutter.

By combining snac’s new media proxying feature with nginx caching, we can achieve a more balanced setup—reducing load on both our instance and external ones, improving privacy, and enhancing performance.