A lot of Warn messages when bootstrapping Tor 0.3.2.9 (stable)
Hello, I'm running a tor-relay with Tor version 0.3.2.9 (stable). I had to do some daemon restarts, because of connection limit issues. Since the last two restarts the Tor process issues a lot of warnings while bootstrapping, because it could not create a (client) circuit. Relay functionality is working without problems. Before these restarts I had no problems connecting to the tor network and creating circuits. Tor usually booted within less than a minute without warnings. I found a ticket describing the same problem (although for a different Tor version) here: https://trac.torproject.org/projects/tor/ticket/24639. Are there any known workarounds yet? For more detail, I'll post the last warning messages: Jan 25 18:31:54.000 [warn] Problem bootstrapping. Stuck at 90%: Establishing a Tor circuit. (Connection timed out; TIMEOUT; count 1503; recommendation warn; host B7047FBDE9C53C39011CA84E5CB2A8E3543066D0 at 199.249.223.60:443) Jan 25 18:31:54.000 [warn] 1503 connections have failed: Jan 25 18:31:54.000 [warn] 1026 connections died in state connect()ing with SSL state (No SSL object) Jan 25 18:31:54.000 [warn] 392 connections died in state handshaking (TLS) with SSL state SSLv2/v3 read server hello A in HANDSHAKE Jan 25 18:31:54.000 [warn] 84 connections died in state handshaking (Tor, v3 handshake) with SSL state SSL negotiation finished successfully in OPEN Jan 25 18:31:54.000 [warn] 1 connections died in state handshaking (TLS) with SSL state unknown state in HANDSHAKE Tor had to try to open a circuit for more than half an hour before finally succeeding... Regards, Dark
On 26 Jan 2018, at 19:55, Dark Matter <tor-relay@gmx.net> wrote:
Hello,
I'm running a tor-relay with Tor version 0.3.2.9 (stable). I had to do some daemon restarts, because of connection limit issues.
There has been extra load on the network since December. We are working on some relay defences: https://lists.torproject.org/pipermail/tor-relays/2018-January/014175.html
Since the last two restarts the Tor process issues a lot of warnings while bootstrapping, because it could not create a (client) circuit. Relay functionality is working without problems.
Your relay has probably chosen some overloaded guards. The problem comes from the extra load on those machines, not your relay. Your relay will eventually try other guards. Are you sure it's not overloaded? It might be worth checking the logs, inbound connections, RAM, and CPU after it has been running for a few days.
Before these restarts I had no problems connecting to the tor network and creating circuits. Tor usually booted within less than a minute without warnings. I found a ticket describing the same problem (although for a different Tor version) here: https://trac.torproject.org/projects/tor/ticket/24639. Are there any known workarounds yet?
Ignore the warning messages unless your relay is having problems. If your relay's traffic drops, or it starts using excessive CPU, RAM, or sockets, there are some things you can do. You can change your tor config, or the number of file descriptors, or block excessive connections from a single IP using a firewall: https://lists.torproject.org/pipermail/tor-relays/2018-January/014100.html
For more detail, I'll post the last warning messages:
Jan 25 18:31:54.000 [warn] Problem bootstrapping. Stuck at 90%: Establishing a Tor circuit. (Connection timed out; TIMEOUT; count 1503; recommendation warn; host B7047FBDE9C53C39011CA84E5CB2A8E3543066D0 at 199.249.223.60:443) Jan 25 18:31:54.000 [warn] 1503 connections have failed: Jan 25 18:31:54.000 [warn] 1026 connections died in state connect()ing with SSL state (No SSL object) Jan 25 18:31:54.000 [warn] 392 connections died in state handshaking (TLS) with SSL state SSLv2/v3 read server hello A in HANDSHAKE Jan 25 18:31:54.000 [warn] 84 connections died in state handshaking (Tor, v3 handshake) with SSL state SSL negotiation finished successfully in OPEN Jan 25 18:31:54.000 [warn] 1 connections died in state handshaking (TLS) with SSL state unknown state in HANDSHAKE
Tor had to try to open a circuit for more than half an hour before finally succeeding...
As long as your relay is fine, there is nothing you can do here. T
On Fri, Jan 26, 2018 at 08:16:20PM +1100, teor wrote:
Your relay has probably chosen some overloaded guards. The problem comes from the extra load on those machines, not your relay. Your relay will eventually try other guards.
Jan 25 18:31:54.000 [warn] Problem bootstrapping. Stuck at 90%: Establishing a Tor circuit. (Connection timed out; TIMEOUT; count 1503; recommendation warn; host B7047FBDE9C53C39011CA84E5CB2A8E3543066D0 at 199.249.223.60:443) Jan 25 18:31:54.000 [warn] 1503 connections have failed: Jan 25 18:31:54.000 [warn] 1026 connections died in state connect()ing with SSL state (No SSL object)
The issue here happens because of two bugs. First is https://bugs.torproject.org/24767 where the flood of circuits causes us to make a flood of connection attempts to relays that are down. And second is https://bugs.torproject.org/projects/25061 (which I just filed) where each of these connection failures gets sent to the "warn about bootstrapping failure" code. So I think these warnings are not actually related to your relay's guard (and that's the bug).
As long as your relay is fine, there is nothing you can do here.
I think this statement is still true. The warnings are harmless. --Roger
participants (3)
-
Dark Matter -
Roger Dingledine -
teor