# Forward caching proxy for a HTTPS server?

**URL:** <https://cernvm-forum.cern.ch/t/forward-caching-proxy-for-a-https-server/244>\
**Category:** Town Square\
**Created:** [October 28, 2022, 11:33pm UTC](https://cernvm-forum.cern.ch/t/forward-caching-proxy-for-a-https-server/244 "2022-10-28T23:33:06Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![rptaylor](https://cernvm-forum.cern.ch/letter_avatar_proxy/v4/letter/r/49beb7/32.png) [@rptaylor](https://cernvm-forum.cern.ch/u/rptaylor)\
**Post date:** [October 28, 2022, 11:33pm UTC](https://cernvm-forum.cern.ch/t/forward-caching-proxy-for-a-https-server/244/1 "2022-10-28T23:33:07Z")

</div>

Suppose we have a CVMFS stratum server (actually S3 object storage) that is only available via HTTPS, not HTTP.  
We want to access it via forward caching proxies as usual, for a CVMFS data repository.

As far as I can tell, the only way to do that involves the “SSL bumping” feature  
[https://wiki.squid-cache.org/Features/SslBump](https://wiki.squid-cache.org/Features/SslBump)  
However that involves security practices we would consider unacceptable.  
We would like to be able to use the object storage for a public repository, working with all the standard caching proxies that people already have set up, preferably without needing much or anything in terms of extra proxy config.

I can envision that the following might be technically possible:  
a client requests [http://s3/datafile](http://s3/datafile) via a proxy only (as DIRECT would not work)  
the proxy has some configuration telling it, for that origin server, it has to connect via HTTPS  
the proxy fetches [https://s3/datafile](https://s3/datafile) , caches the content, and returns the HTTP response (TLS headers etc. stripped off) to the client

However I am not sure if any functionality exists like that in squid, at least I could not find any information about it. @dwd ?  
If it is, it would be an option for forward or reverse proxy caching.

Of course if we had a frontend or reverse proxy to the object storage that was HTTP, that would take care of that part of the problem, but it might be a bottleneck or be less resilient compared to the object storage cluster itself.

Thanks!

---

<div class="post-metadata">

**Author:** ![dwd](https://cernvm-forum.cern.ch/letter_avatar_proxy/v4/letter/d/b9bd4f/32.png) [@dwd](https://cernvm-forum.cern.ch/u/dwd)\
**Post date:** [October 31, 2022, 3:59pm UTC](https://cernvm-forum.cern.ch/t/forward-caching-proxy-for-a-https-server/244/2 "2022-10-31T15:59:59Z")

</div>

Hi Ryan,

I haven’t tried doing what you want to do, but perhaps a `url_rewrite_program` could be used for that. I know people who have successfully used that for other purposes.

Dave
