Opened 6 years ago

Closed 5 years ago

#8898 closed defect (duplicate)

Tor says it closes socks listener, but doesn't actually close it.

Reported by: bastik Owned by: chiiph
Priority: Medium Milestone:
Component: Archived/Vidalia Version: Tor: unspecified
Severity: Keywords: tor-client
Cc: Actual Points:
Parent ID: Points:
Reviewer: Sponsor:

Description

With Tor 0.2.4.12-alpha on Windows 7 64 bit (probably not limited to this set-up) Tor says it would close a socks listener, but it does not. (Lucky me)

I added "SocksPort 127.0.0.1:9090" to my torrc which contained a similar line with port 9001.

Because the instance is a bridge, I didn't want to close and restart Tor. I went to the advanced settings of Vidalia and clicked on edit the current torrc. Then I marked only the newly added line and selected "Apply selected only".

Vidalia logged:
[Notice] Opening Socks listener on 127.0.0.1:9090
[Notice] Closing no-longer-configured Socks listener on 127.0.0.1:9001
[Notice] Closing old Socks listener on 127.0.0.1:9001

but port 9001 is still used by Tor and the connection is not broken.

When I look in the torrc now the line "SocksPort 127.0.0.1:9001" is gone, the "SocksPort 127.0.0.1:9090" remains.

(The commented PublishServerDescriptor 0 is gone as well)

How to reproduce:

  • Launch Tor with SocksPort set to some port
  • Edit your torrc and add another SocksPort line
  • Open Vidalia and apply only the line you added
  • See what appears in the log and how it mismatches what actually happens

Child Tickets

Change History (7)

comment:1 Changed 6 years ago by nickm

Milestone: Tor: 0.2.4.x-final

I have a hunch that if we test this with Tor alone, and not with Vidalia, it will turn out not to actually be a Tor bug. Anybody got a chance to try it out?

comment:2 in reply to:  1 Changed 6 years ago by bastik

Replying to nickm:

I have a hunch that if we test this with Tor alone, and not with Vidalia, it will turn out not to actually be a Tor bug. Anybody got a chance to try it out?

I guess you are true.

Not sure if I can test this as I don't think Windows supports sending SIGHUP (or any other signal) to a process. (please tell me otherwise).

It seems to turn out that when I apply two socks lines with Vidalia only the bottom one survives. If I edit the torrc (re-adding the lost line) and restart Tor it binds to two socks ports. So maybe just one bug in Vidalia.

comment:3 Changed 6 years ago by arma

sounds like the vidalia bug where it setconfs the lines in the torrc text box one at a time.

we should really disable that whole vidalia torrc edit box until somebody starts maintaining vidalia again.

comment:4 Changed 6 years ago by arma

comment:5 Changed 6 years ago by arma

oh, and #6147 and #7535 and #7718 and #7719.

I think #3927 and #8893 are different though.

comment:6 Changed 6 years ago by nickm

Component: TorVidalia
Milestone: Tor: 0.2.4.x-final
Owner: set to chiiph

comment:7 Changed 5 years ago by bastik

Resolution: duplicate
Status: newclosed

Back then I cared about the mismatch between what the log said and what happened. And it turned out to be duplicate of #5901. Closing as duplicate as I think this ticket is just that bug. Plus Vidalia is unmaintained.

Note: See TracTickets for help on using tickets.