Opened 5 years ago

Last modified 5 months ago

#10510 new defect

Getting a lot of system clock jumped messages

Reported by: LoneRanger1012 Owned by:
Priority: Medium Milestone: Tor: unspecified
Component: Core Tor/Tor Version: Tor: 0.2.4.18-rc
Severity: Normal Keywords: tor-relay, tor-client, sponsor8-maybe, timekeeping, wakeup, 032-unreached
Cc: Actual Points:
Parent ID: Points:
Reviewer: Sponsor:

Description

Your help desk suggested that I create this ticket.
I am running Tor 0.2.4.18-rc (git-1cda452bc136de6b) on my 64 bit Windows
7 machine. It is configured as a non-exit relay with the nick name
4U2Guess.
Since updating to the 2.4 series several months ago I have been getting
lots of system clock jumped messages, often several per day, and
sometimes for intervals as long as 16105 seconds (4-1/2 hrs). As
expected, the program starts over building new circuits, which takes a
long time and reduces my average throughput dramatically. After that
4-1/2 hr gap it took 11 minutes to generate a message that Tor had
finally successfully opened a circuit

I think that the problem possibly happens because some of the Tor
program code 'hogs' the part of the cpu time allocated to Tor and
Vidalia effectively preventing the part of the program that checks the
clock from executing, sometimes for hours. While these gaps usually
fall while I am away from the console and the screen saver has shut off
the monitor, I do not think that is the cause, but I am not sure how to
test that.

I usually have the microsoft Resource and Performance Monitor
(perfmon.exe) running. While it shows a lot of interesting stuff, it
does not show things that I am interpreting as evidence that Vidalia or
Tor are misbehaving or being adversely affected.

Some of my system details reported by dxdiag:

Time of this report: 12/28/2013, 08:13:12

Machine name: DLC-THINKCENTRE

Operating System: Windows 7 Professional 64-bit (6.1, Build 7601)

Service Pack 1 (7601.win7sp1_gdr.130828-1532)

Language: English (Regional Setting: English)

System Manufacturer: LENOVO

System Model: 3133A9U

BIOS: LENOVO BIOS Rev: 1.0

Processor: Intel(R) Pentium(R) CPU G630 @ 2.70GHz (2 CPUs),

~2.7GHz

Memory: 6144MB RAM

Available OS Memory: 5992MB RAM

Page File: 5092MB used, 6890MB available

Windows Dir: C:\Windows

DirectX Version: DirectX 11

DX Setup Parameters: Not found

User DPI Setting: 96 DPI (100 percent)

System DPI Setting: 96 DPI (100 percent)

DWM DPI Scaling: Disabled

DxDiag Version: 6.01.7601.17514 32bit Unicode

Do you have some suggestions about how to track this problem down?

Thanks.

David C

Child Tickets

Attachments (1)

VidaliaLog-12.28.2013.txt (4.8 KB) - added by LoneRanger1012 5 years ago.
Vidalia log - the 4-1/2 hr gap is near the middle

Download all attachments as: .zip

Change History (18)

Changed 5 years ago by LoneRanger1012

Attachment: VidaliaLog-12.28.2013.txt added

Vidalia log - the 4-1/2 hr gap is near the middle

comment:1 Changed 5 years ago by LoneRanger1012

I should have mentioned that I have the AT&T branded version of McAfee anti-virus running on this machine. I have it set to ask me before updating, and I normally allow it within a few minutes or hours.

The real time scanning feature sometimes slows the entire machine down. I have tried to configure it to leave Tor and Vidalia alone, but it still sometimes blocks Tor for some reason.

comment:2 Changed 5 years ago by cypherpunks

While these gaps usually fall while I am away from the console and the screen saver has shut off the monitor, I do not think that is the cause, but I am not sure how to test that.

Are you sure your OS is not configured for sleep mode on user inactivity? That could explain so much seconds of clock jumping.

comment:3 Changed 5 years ago by cypherpunks

Try to bring Insomnia for your box without any changes of OS settings. It still should allow sleep of monitor, but keep CPU awake.

comment:4 Changed 5 years ago by LoneRanger1012

  1. I get a 500 Server error when I try to download that Insomnia program.
  2. There are a few reasons that I think that the computer is not going to sleep. One very strong one is that Vidalia continues entering messages in the log. But then I do not know what events can wake it up besides keyboard activity.

comment:5 Changed 5 years ago by cypherpunks

Vidalia continues entering messages in the log.

Actually no. Just tested several clock jumps from attached log, Vidalia didn't wrote any messages for more than seconds of clock jump before message with about clock jump.
Like:

Dec 28 02:13:58.022 [Notice] Circuit handshake stats since last time: 415/415 TAP, 4/4 NTor.
17
Dec 28 07:17:48.733 [Warning] Your system clock just jumped 16105 seconds forward; assuming established circuits no longer work.

Dec 28 07:17:48.733 - Dec 28 02:13:58.022 = ~ 303 minutes
16105 seconds = ~ 268 minutes.
Vidalia didn't wrote anything while CPUs (probably) was asleep.

comment:6 Changed 5 years ago by cypherpunks

Task Idle Conditions so complex, and modern box so smart. CPU probably still counts but some part of mainboard was turned-off for preserve energy. Need investigation, Tor deserve fix for such conditions.

comment:7 Changed 5 years ago by cypherpunks

Component: - Select a componentTor

comment:8 Changed 5 years ago by LoneRanger1012

I just found the System event log. For the 16105 second gap at 7:17 AM December 28 the system log reports that the computer went to sleep at 2:49:02 AM and woke at 7:17:29 AM.

Is there some way for Vidalia or Tor to tell the system that they are busy? Or to figure out why the computer concluded that it was idle and went to sleep?

comment:9 Changed 5 years ago by LoneRanger1012

When the computer woke up it reported the reason was because the USB Root hub woke it. As you say, it is complex. I must have failed to set something in the power monitor, possibly something that is not even visible to any but the most savvy user. If that is the case, I would like to have Tor do something to help me prevent this problem. Even a note to check certain computer settings would help.

comment:10 Changed 5 years ago by cypherpunks

Seems like Tor have no peaks of CPU usage, and using so small resources that Windows count system like idling and goes to sleep. You can to change power save settings to prevent sleep if runs Tor as relay.

Tor could to use SetThreadExecutionState API to notify Windows about major mission for relay case and/or by torrc option (with default allowing to sleep). It's simple but platform specific stuff.

Last edited 5 years ago by cypherpunks (previous) (diff)

comment:11 Changed 5 years ago by LoneRanger1012

At some time early in the lifecycle of each new computer that I get, I set certain parameters such as sleep options and monitor refresh rates to my personal preferences, yet later I find that they have somehow changed. If the monitor refresh rate is less than 70 hertz, I get headaches. Yet, from time to time it somehow changes back to 60 Hertz. Apparently, somehow the computer got set to sleep after 40 minutes, even though I am sure that when I first started running Tor I set it to never sleep, but to allow the monitor to be turned off. Apparently, by some unknown coincidence, it changed at about the same time that I upgraded to the Tor 0.2.4 alpha series, as that is about the time that I started seeing the clock messages. I do not normally re-check settings that supposedly do not change magically. When I re-checked the advanced power settings I did not see a way to set it to prevent sleep if, for example, Tor is running as a relay.

Are you saying that I could add a line to my torrc to accomplish that?

comment:12 Changed 5 years ago by cypherpunks

Are you saying that I could add a line to my torrc to accomplish that?

Not with current code, but if Tor could implement it in the future version. Not sure what Tor developers think about it, however.

comment:13 Changed 5 years ago by nickm

Milestone: Tor: unspecified

comment:14 Changed 2 years ago by nickm

Keywords: tor-relay tor-client sponsor8-maybe timekeeping wakeup added
Severity: Normal

comment:15 Changed 2 years ago by nickm

Milestone: Tor: unspecifiedTor: 0.3.2.x-final
Sponsor: Sponsor8-can

comment:16 Changed 21 months ago by nickm

Keywords: 032-unreached added
Milestone: Tor: 0.3.2.x-finalTor: unspecified

Mark a large number of tickets that I do not think we will do for 0.3.2.

comment:17 Changed 5 months ago by gaba

Sponsor: Sponsor8-can
Note: See TracTickets for help on using tickets.