Thanks for reporting this issue, Rick.
On Wed, Jun 19, 2019 at 23:30:40 UTC, Rick Huebner wrote:
It looks like bridge 1 (eldritchworld) gave me bad/incomplete relay info which prevented making circuits [...]
For what it's worth, I managed to bootstrap an obfs4 connection over this bridge.
I know the new bridgedb release doesn't give out ORports if there's a secure PT offered, but last I checked the majority of bridges were just vanilla, so surely there must be enough to put some into whichever bridge pool ring my request was pulled from.
Right now, 1,018 bridges have the 'Running' flag. 585 (57.5%) of these 1,018 bridges support obfs4.
Could someone else check if they see similar problems?
We're aware of this problem. See the following ticket for (slightly) more information: https://bugs.torproject.org/30441
As far as I can tell, there are at least three problems:
1. We don't have enough bridges and after fixing #28655, we are left with even fewer vanilla bridges.
2. Some bridges have their obfs4 port firewalled. After reaching out, some operators fixed this issue. Other obfs4 bridges don't have contact info, so we will have to blacklist them.
3. There's some churn among bridges and once a bridge goes offline, it takes ~2-3 hours until this event propagates to BridgeDB. In the meanwhile, BridgeDB is still handing out the bridge line.
I wouldn't be surprised if there's another issue related to how BridgeDB allocates bridges into rings. I'll look into it.
On a general note, BridgeDB development has slowed down a bit because a long-time maintainer is not around anymore and its current maintainers (which includes me) are just getting up to speed.
Cheers, Philipp