Re: server incorrectly believes IP address has changed

[Scott Bennett <bennett@xxxxxxxxxx>]

     On Tue, 20 Nov 2007 20:37:25 +0000 Robert Hogan <robert@xxxxxxxxxxxxxxx>
wrote:
>On Tuesday 20 November 2007 19:52:43 you wrote:
>>      On Tue, 20 Nov 2007 16:59:24 +0100 Csaba Kiraly <kiraly@xxxxxxxxxxxx>
>>
>> wrote:
>> >If you did not specify your IP within the torrc, tor tries to guess it
>> >based on your hostname ... and this goes wrong if the DNS is wrong!
>>
>>      As stated in the text you top-posted, if you had read it, I specified
>> a host+domain name in torrc.  That host+domain name combination is updated
>> at dyndns.org upon system startup and within ten minutes of any change.
>> I use inadyn to test and update the dyndns.org name server data base.
>>
>
>But you didn't specify your IP in the torrc, so Csaba's point is still a=20
>possibility.

     Actually, according to the documentation, the guessing he referred to
happens when "Address" is not provided in torrc, which is quite a different
matter from specifying an Address in host+domain name form.
>
>However, since you didn't delete cached-routers before start-up on either=20

     cached-routers and cached-routers.new are not the file names used in
0.2.0.12-alpha, but rather cached-descriptors and cached-descriptors.new, and
in my original message, I wrote:

 ->     I restarted my tor server a couple of hours ago, and since then, it has
 ->been acting very peculiarly.  Here are the notice-level log file entries since
 ->startup.  Note that I deleted cached-descriptors and cached-descriptors.new
 ->after shutting down tor and before shutting down the system several hours prior
 ->to this startup.

>occasion my own *guess* would be that tor is using the old IP stored there,=
> =20
>especially since 66.225.42.30 was your address at one point:
>=20
>http://www.google.com/search?hl=3Den&q=3Dmycroftsotherchild+66.225.42.30&bt=
>nG=3DSearch

     Yes, the ISP has a limited list of IP addresses that it assigns, so the
same addresses do recur frequently.  In this case, I noted, 

 ->Nov 20 06:29:37.282 [notice] Now checking whether ORPort 66.225.36.5:995 and DirPort 66.225.36.5:443 are reachable... (this may take up to 20 minutes -- look for log messages indicating success)
 ->
 ->     Note that the above address was incorrect.  The correct address was, and
 ->still is, 66.225.42.30.  It has not changed since before the system was
 ->rebooted.

>
>Whether it 's a good thing for Tor to just try the last known good address=
>=20
>rather than  figure it out all over again on the off-chance it may be out o=
>f=20
>date, I don't know.
>
     I thought the point of specifying a host+domain name in the "Address "
line was to get tor to *look it up in the name server net*.
>
>The tone of your comments below is unfair to Csaba. He was merely offering =
>a=20
>friendly (and free) suggestion. No one would ever respond to anyone else on=
>=20
>this list if that sort of thing was the default response.

     In that case, I apologize.  I took offense at the comments that appeared
to ignore what I had written, while taking a very condescending tone and
assuming I hadn't done the obvious checking before posting.
>
>> >I think messages related to this are at info level, so you can have a
>> >look at info level logs.
>> >
>> >If you are running on Linux, look at your /etc/hosts file.
>>
>>      I run FreeBSD 6-STABLE.  There is no entry in /etc/hosts because
>> /etc/nsswitch.conf is set to have the resolver routines check /etc/hosts
>> before querying name servers.  An entry in /etc/hosts for the chosen
>> host+domain name combination at dyndns.org would defeat the purpose of
>> using inadyn and the dyndns.org service.  Although I have not spent time
>> to familiarize myself with the tor source code, especially since I have
>> yet to encounter written documentation of its internal structure, I am
>> not an idiot.  I have also been doing UNIX system administration for two
>> decades.
>>
>> >Check also what you get from DNS lookup for your hostname with
>> >"nslookup" (both Linux and Windows).
>>
>>      nslookup is a pain in the butt.  I normally use dig.  In any case,
>> as stated in the text that you ignored, the IP address had not changed.
>> Therefore it had not been altered at dyndns.org.  And, yes, I did check.
>> The address had not changed at dyndns.org, and it had not changed here.
>> The ADSL modem+router's log showed only the initial PPPoE exchange that
>> resulted from plugging it in.  If the IP address had changed, all of that
>> would have had to appear in the modem+router's log.
>>      tor has now been up and running for just over five hours since I
>> stopped and restarted it because of the problem.  There has been no
>> recurrence during this time.

     And it is still running with no sign of trouble.  However, it occurs
to me that when I stopped tor and restarted it in hopes of getting rid of
the bad behavior, I neglected to re-delete cached-descriptors and
cached-descriptors.new, so perhaps the aberrant activity is related to
starting up and finding no cached-descriptors* files.  I have often deleted
them before starting tor in order to reduce the volume of long-obsoleted
descriptors.  tor doesn't seem to replace the old ones with the new ones,
but rather tends to accumulate a large number of obsolete descriptors before
eventually partially cleaning up the files.


                                  Scott Bennett, Comm. ASMELG, CFIAG
**********************************************************************
* Internet:       bennett at cs.niu.edu                              *
*--------------------------------------------------------------------*
* "A well regulated and disciplined militia, is at all times a good  *
* objection to the introduction of that bane of all free governments *
* -- a standing army."                                               *
*    -- Gov. John Hancock, New York Journal, 28 January 1790         *
**********************************************************************