Exit Policy says accept *:* but relay doesn't have Exit flag

#13718 (moved) has revealed an issue where chutney-run tor authorities don't flag anything as an Exit. This was resolved in #13161 (moved) with TestingDirAuthVoteExit, but a more elegant solution (or a root cause) needs to be found in order to properly test #13718 (moved).

Workarounds are available: TestingDirAuthVoteExit * AssumeReachable 0 (not actually a workaround)

But they make it harder to justify that the test results apply to real-world tor bootstraps, or even chutney configs without special-case configurations.

This issue is like #11264 (moved), and could have the same root cause (a conflict between the setting of the exit flag and exit policy summaries), although this seems unlikely.

added component::core tor/tor owner::teor parent::13718 priority::medium resolution::fixed status::closed tor-auth tor-relay type::defect version::tor 0.2.6.1-alpha labels

Trac:
torrc

chutney-generated torrc for an authority which doesn't label itself Exit even though it has accept :

Typo in description

Trac:
Description: #13718 (moved) has revealed an issue here chutney-run tor authorities don't flag anything as an Exit. This was resolved in #13161 (moved) with TestingDirAuthVoteExit, but a more elegant solution (or a root cause) needs to be found in order to properly test #13718 (moved).

Workarounds are available: TestingDirAuthVoteExit * AssumeReachable 0

But they make it harder to justify that the test results apply to real-world tor bootstraps, or even chutney configs without special-case configurations.

This issue is like #11264 (moved), and could have the same root cause (a conflict between the setting of the exit flag and exit policy summaries), although this seems unlikely.

to

#13718 (moved) has revealed an issue where chutney-run tor authorities don't flag anything as an Exit. This was resolved in #13161 (moved) with TestingDirAuthVoteExit, but a more elegant solution (or a root cause) needs to be found in order to properly test #13718 (moved).

Workarounds are available: TestingDirAuthVoteExit * AssumeReachable 0

But they make it harder to justify that the test results apply to real-world tor bootstraps, or even chutney configs without special-case configurations.

This issue is like #11264 (moved), and could have the same root cause (a conflict between the setting of the exit flag and exit policy summaries), although this seems unlikely.

AssumeReachable 0 is not actually a workaround. TestingDirAuthVoteExit * from #13161 (moved) is the only available workaround for this issue.

Trac:
Keywords: tor-auth deleted, tor-auth lorax added

Trac:
Description: #13718 (moved) has revealed an issue where chutney-run tor authorities don't flag anything as an Exit. This was resolved in #13161 (moved) with TestingDirAuthVoteExit, but a more elegant solution (or a root cause) needs to be found in order to properly test #13718 (moved).

Workarounds are available: TestingDirAuthVoteExit * AssumeReachable 0

But they make it harder to justify that the test results apply to real-world tor bootstraps, or even chutney configs without special-case configurations.

This issue is like #11264 (moved), and could have the same root cause (a conflict between the setting of the exit flag and exit policy summaries), although this seems unlikely.

to

#13718 (moved) has revealed an issue where chutney-run tor authorities don't flag anything as an Exit. This was resolved in #13161 (moved) with TestingDirAuthVoteExit, but a more elegant solution (or a root cause) needs to be found in order to properly test #13718 (moved).

Workarounds are available: TestingDirAuthVoteExit * AssumeReachable 0 (not actually a workaround)

But they make it harder to justify that the test results apply to real-world tor bootstraps, or even chutney configs without special-case configurations.

This issue is like #11264 (moved), and could have the same root cause (a conflict between the setting of the exit flag and exit policy summaries), although this seems unlikely.
Status: new to assigned
Owner: N/A to teor

I think this issue occurs because dirserv_compute_performance_thresholds() assigns exit flags after filtering by router_counts_toward_thresholds(). This appears to be a feature intended to avoid pollution of the Tor network by low-quality Exit nodes.

The fix is to use an option that's already in the code. I think TestingMinExitFlagThreshold 0 should ensure the exit flag is assigned to all potential exit relays, after they self-test reachability and post their descriptors to the authorities.

Trac:
Keywords: tor-auth lorax deleted, tor-auth tor-relay added

A patch is required to make TestingMinExitFlagThreshold 0 work the way I assume it should. I expect that TestingMinExitFlagThreshold 0 would cause authorities to ignore bandwidth (either advertised or measured) when assigning the Exit flag.

But router_is_active() marks routers as inactive if they have no bandwidthcapacity in their routerinfo.

My suggested patch is to ignore the non-zero bandwidthcapacity requirement when TestingTorNetwork is set, and TestingMinExitFlagThreshold 0.

This causes the authorities to mark themselves, and any qualifying relays, as Exits.

The changes to tor and chutney in #13718 (moved) have fixed this:

Bugs: #13718 (moved), #13814 (moved), maybe #13787 (moved), #13839 (moved), #13924 (moved), #13823 (moved), #13929 (moved), #13963 (moved) Branch: bug13718-fast-bootstrap Note: There are 5 branches that start with bug13718, please choose the right one. Repository: ​​​​​​​​https://github.com/teor2345/tor.git

Bugs: #13823 (moved) Branch: bug13823-fast-bootstrap Repository: ​​​​​​​​https://github.com/teor2345/chutney.git

Trac:
Status: assigned to needs_review

tor changes committed as part of of bug13718-consensus-interval merge. chutney changes have not yet been merged.

dgoulet has tested the chutney changes along with the draft tor changes in #13718 (moved).

With the merge of the chutney and torspec changes in #13718 (moved), this issue is now resolved.

Trac:
Resolution: N/A to fixed
Status: needs_review to closed

closed

mentioned in issue #13924 (moved)

mentioned in issue #13928 (moved)

mentioned in issue #13929 (moved)

mentioned in issue #13934 (moved)

mentioned in issue #13935 (moved)

mentioned in issue #13963 (moved)

moved to tpo/core/tor#13839 (closed)