strange consensus messages in log
Yesterday built openssl-1.0.1m and dropped it over the top of 1.0.1j. Tor said OpenSSL version from headers does not match the version we're running with. If you get weird crashes, that might be why. (Compiled with 100010af: OpenSSL 1.0.1j 15 Oct 2014; running with 100010df: OpenSSL 1.0.1m 19 Mar 2015). Can't see why this would cause the strange consensus error (identical ABI, right?), but just in case rebuilt tor relay 0.2.4.26 against the rebuilt openssl and restarted. At 22:32 4/5/2015 -0400, you wrote:
Does anyone have any idea what this is about?
Is it serious?
Hi, Very strange ! Data have been corrupted (even the hour), but it seems to be repeated and constant.
From my experiences, every "newer library" contains the "older library" functionalities (even if improved), it's supposed to be the goal of a library. Can somebody confirm that about OpenSSL ?
Data could have been corrupted : -> Before reception (strange, it seems to come from several source) - unlikely I would said, but it's not a truth, just my opinion. -> After reception (strange also : what is the cause ?) Another remark : Unable to load microdesc consensus directory downloaded from server '194.109.206.212:80'. I'll try again soon. This maybe doesn't need OpenSSL ? Generally, defective memory modules are driving to random errors executing programs (those are located (or working with data located) at the faulty emplacement), and with Windows Kernels, it drives to OS crashes and reboots everywhere (may be it could be the same on Linux ?). But you can check with memtest (http://www.memtest.org/) started in place of your OS. If it's "ECC" memory module, a slightly defective ECC memory module cannot be the cause of a "successful data corruption" (the data is saved from corruption in extremis, and you should be warned - but I don't know how). Keep us informed ! And good luck. Best regards, Julien ROBIN ----- Original mail ----- From: "starlight 2015q2" <starlight.2015q2@binnacle.cx> To: tor-relays@lists.torproject.org Sent: Lundi 6 Avril 2015 06:55:17 Object: Re: [tor-relays] strange consensus messages in log Yesterday built openssl-1.0.1m and dropped it over the top of 1.0.1j. Tor said OpenSSL version from headers does not match the version we're running with. If you get weird crashes, that might be why. (Compiled with 100010af: OpenSSL 1.0.1j 15 Oct 2014; running with 100010df: OpenSSL 1.0.1m 19 Mar 2015). Can't see why this would cause the strange consensus error (identical ABI, right?), but just in case rebuilt tor relay 0.2.4.26 against the rebuilt openssl and restarted. At 22:32 4/5/2015 -0400, you wrote:
Does anyone have any idea what this is about?
Is it serious?
_______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
PS : I think a defective network equipment could also be the cause of such data corruptions on incoming data ? May be it should be added on the check list ! But if it's the case, it should be the case for every kind of connexion (try to download an unzip some big file, just to be sure ;) ----- Mail original ----- De: "Julien ROBIN" <julien.robin28@free.fr> À: tor-relays@lists.torproject.org Envoyé: Lundi 6 Avril 2015 13:54:38 Objet: Re: [tor-relays] strange consensus messages in log Hi, Very strange ! Data have been corrupted (even the hour), but it seems to be repeated and constant.
From my experiences, every "newer library" contains the "older library" functionalities (even if improved), it's supposed to be the goal of a library. Can somebody confirm that about OpenSSL ?
Data could have been corrupted : -> Before reception (strange, it seems to come from several source) - unlikely I would said, but it's not a truth, just my opinion. -> After reception (strange also : what is the cause ?) Another remark : Unable to load microdesc consensus directory downloaded from server '194.109.206.212:80'. I'll try again soon. This maybe doesn't need OpenSSL ? Generally, defective memory modules are driving to random errors executing programs (those are located (or working with data located) at the faulty emplacement), and with Windows Kernels, it drives to OS crashes and reboots everywhere (may be it could be the same on Linux ?). But you can check with memtest (http://www.memtest.org/) started in place of your OS. If it's "ECC" memory module, a slightly defective ECC memory module cannot be the cause of a "successful data corruption" (the data is saved from corruption in extremis, and you should be warned - but I don't know how). Keep us informed ! And good luck. Best regards, Julien ROBIN ----- Original mail ----- From: "starlight 2015q2" <starlight.2015q2@binnacle.cx> To: tor-relays@lists.torproject.org Sent: Lundi 6 Avril 2015 06:55:17 Object: Re: [tor-relays] strange consensus messages in log Yesterday built openssl-1.0.1m and dropped it over the top of 1.0.1j. Tor said OpenSSL version from headers does not match the version we're running with. If you get weird crashes, that might be why. (Compiled with 100010af: OpenSSL 1.0.1j 15 Oct 2014; running with 100010df: OpenSSL 1.0.1m 19 Mar 2015). Can't see why this would cause the strange consensus error (identical ABI, right?), but just in case rebuilt tor relay 0.2.4.26 against the rebuilt openssl and restarted. At 22:32 4/5/2015 -0400, you wrote:
Does anyone have any idea what this is about?
Is it serious?
_______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays _______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
participants (2)
-
Julien ROBIN -
starlight.2015q2＠binnacle.cx