We currently have around 500 obfs4 bridges. That's not a lot and some of them don't work properly, so the effective number is even lower. It's time to ask volunteers to set up more bridges.
One way to do that is by writing a blog post and doing outreach on Twitter. We should point people to our obfs4 setup guide. Ideally, we would start with a small group of people (e.g., by posting to tor-relays@), fix any issues that we discover in our guide, and then repeat the outreach campaign for a larger set of people.
Another thing worth considering is what kind of bridges we want:
We certainly need more obfs4 bridges even though the protocol still has some UX issues for operators (see #30471 (moved)).
More vanilla bridges would be useful too but, since #28655 (moved), only if the bridge doesn't also run an obfs4 protocol. Otherwise BridgeDB won't hand out the vanilla bridge line. As an added bonus, modern tor bridges may currently be resistant to the GFW's active probing as discussed over in #30500 (moved).
Once #23888 (moved) is done, we should also encourage people to install the snowflake webextension.
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information
Child items
0
Show closed items
No child items are currently assigned. Use child items to break down this issue into smaller parts.
Linked items
0
Link issues together to show that they're related.
Learn more.
About this, we will include obfs4 guide to the new Community Portal section "Relay Operations", so we can translate and it will be easier to find and share this resource. The new portal is going to be publicly released till the end of June.
Also, should we have a goal in numbers? How many more bridges we are looking for? We can discuss this during a community team meeting.
About this, we will include obfs4 guide to the new Community Portal section "Relay Operations", so we can translate and it will be easier to find and share this resource. The new portal is going to be publicly released till the end of June.
That's great news, thanks!
Also, should we have a goal in numbers? How many more bridges we are looking for? We can discuss this during a community team meeting.
In our proposal for Sponsor 30 we wrote that "we aim to attract volunteers to run at least 100 new bridges."
There is a ticket on Stack Exchange by someone who says, "I am running an obfs4 bridge on my raspberry pi since 10 days now and haven't seen a single client," and a reply that says, "I have the same problem."
Over at #30708 (moved), I started working on a Docker image for obfs4proxy (and Tor). Some prospective bridge operators may find this useful, so we may want to get this done before launching our campaign.
Sounds good. Adding #30708 (moved) as a ticket then and we come back here once that is done.
We're done with the docker container and added instructions to our setup guide. We can now move forward with this ticket.
Actually, let's wait until #31153 (moved) is done as it will make bridge setup significantly easier. We hope to have the package in Debian stable by mid-September.
Hi! We have a roadmap item on the metrics team roadmap saying "Be prepared to track the "set up a new {relay,bridge}" campaigns (c.f. #30777 (moved))".
Are there any more details about what kind of preparation is needed here? Who is going to track this campaign, and what information do they need? Would a Relay Search query like this one (with whatever number of days back) help as first insight into newly added bridges? Or is this about making fancy graphs at the end of the campaign and maybe mid-way through?
Happy to discuss this on a new ticket or elsewhere. This just seemed like a way to reach everyone involved. Sorry for the noise!
Hi! We have a roadmap item on the metrics team roadmap saying "Be prepared to track the "set up a new {relay,bridge}" campaigns (c.f. #30777 (moved))".
Are there any more details about what kind of preparation is needed here? Who is going to track this campaign, and what information do they need? Would a Relay Search query like this one (with whatever number of days back) help as first insight into newly added bridges? Or is this about making fancy graphs at the end of the campaign and maybe mid-way through?
Happy to discuss this on a new ticket or elsewhere. This just seemed like a way to reach everyone involved. Sorry for the noise!
We have a DRL objective in which we mentioned that we aim for at least 100 new bridges. Naturally, it's difficult to show that a bridge was set up only because of our campaign. (While our bridge setup instructions have hard-coded nicknames that could help us approximate the number of "campaign bridges," we encourage users to pick whatever nickname they like.) Your Relay Search query is indeed helpful: the metric "number of new bridges since the campaign started" is probably the best approximation we have to track our success.
This falls under our sponsor 30 deliverable "Improve documentation on how to set up a bridge server and different pluggable transport bridge servers." We spent a lot of time publishing our bridge setup instructions.
Our campaign ended on September 30th. Here's a summary:
Volunteers set up 96 new bridges – mostly running obfs4, but some run vanilla Tor.
We randomly picked 10 winners to receive a t-shirt.
I manually tested all new bridges. Most worked right away but several used obsolete Tor versions that were no longer accepted by Serge, some did not have accessible obfs4 ports, and a few operators set up relays instead of bridges.
A volunteer generously offered to set up dozens of high-performance private bridges for manual distribution.
Several volunteers pointed out issues in existing setup guides, contributed new guides, and suggested improvements for our docker image (#31834 (moved)).
Here's a visualisation of our bridge growth. Note that we launched our campaign blog post on August 28:
/Edit: Some concluding thoughts for the next time we are running a contest:
The t-shirts were a great incentive. Every other volunteer mentioned that they would really like to win one. We haven't gotten a single email since the contest ended, and there were no more t-shirts to win.
Some volunteers struggled with setting up a bridge and asked for ways to test their bridge themselves. It would have helped if #31874 (moved) was done before we started the campaign. Regardless, many volunteers were excited to get a personalised email in response to their submission, so we shouldn't automate everything.
We should have mentioned what Tor versions were obsolete and no longer accepted by Serge. Several operators installed obsolete Tor versions from their package repository.
Ask volunteers to provide us with their bridge line, instead of their hashed fingerprint. This would make it easier for us to test bridges.