Unusual TSIG problem
I just ran into an odd issue with a TSIG signed zone transfer. On occasion I was logging a clocks are unsynchronized message doing a transfer from a customer server at a site about 30 ms away. I dropped a note to the manager there asking that he look at the his system for a time issue. He checked and found no problems. Today I looked at the problem more closely. I realized that the problem was NOT a clock sync issue. They were probably within a millisecond of one another. I found the following in the log: Dec 8 06:26:18 ns1 named[67170]: zone XX.gov/IN: notify from 123.234.1.1#33372: refresh in progress, refresh check queued Dec 8 06:31:18 ns1 named[67170]: transfer of 'XX.gov/IN' from 123.234.1.1#53: failed while receiving responses: clocks are unsynchronized Dec 8 06:31:18 ns1 named[67170]: transfer of 'XX.gov/IN' from 123.234.1.1#53: Transfer completed: 1 messages, 397 records, 59674 bytes, 898.462 secs (66 bytes/sec) The transfer, probably due to a hardware problem was taking over 5 minutes to transfer the zone and RFC2845 suggests tha the difference between clocks should be limited to 300 seconds (5 minutes). This really means that, should the transfer take over 5 minutes, you get the unsynced clocks error. (4.5.2. TIME check and error handling) Clearly, something is broken when a zone transfer takes over 5 minutes. (This one SHOULD take about 2-3 seconds.) But the message certainly pointed in the wrong direction. Is there more appropriate language that might indicate that it could also be an effective time-out because the transfer took too long? Maybe "failed while receiving responses: clocks are unsynchronized or maximum transfer time exceeded"? -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: ober...@es.net Phone: +1 510 486-8634 Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751 ___ bind-users mailing list bind-users@lists.isc.org https://lists.isc.org/mailman/listinfo/bind-users
Re: Unusual TSIG problem
In message <20101208214221.566771c...@ptavv.es.net>, "Kevin Oberman" writes: > I just ran into an odd issue with a TSIG signed zone transfer. > > On occasion I was logging a clocks are unsynchronized message doing a > transfer from a customer server at a site about 30 ms away. I dropped a > note to the manager there asking that he look at the his system for a > time issue. He checked and found no problems. > > Today I looked at the problem more closely. I realized that the problem > was NOT a clock sync issue. They were probably within a millisecond of > one another. I found the following in the log: > Dec 8 06:26:18 ns1 named[67170]: zone XX.gov/IN: notify from 123.234.1.1 > #33372: refresh in progress, refresh check queued > Dec 8 06:31:18 ns1 named[67170]: transfer of 'XX.gov/IN' from 123.234.1. > 1#53: failed while receiving responses: clocks are unsynchronized > Dec 8 06:31:18 ns1 named[67170]: transfer of 'XX.gov/IN' from 123.234.1. > 1#53: Transfer completed: 1 messages, 397 records, 59674 bytes, 898.462 secs > (66 bytes/sec) > > The transfer, probably due to a hardware problem was taking over 5 > minutes to transfer the zone and RFC2845 suggests tha the difference > between clocks should be limited to 300 seconds (5 minutes). This really > means that, should the transfer take over 5 minutes, you get the > unsynced clocks error. (4.5.2. TIME check and error handling) 59674*8/397 = 1202 b/s that's slower than almost all dialup lines. If this happens regularly use transfer-format multiple-messages which results in smaller messages and more signatures. > Clearly, something is broken when a zone transfer takes over 5 > minutes. (This one SHOULD take about 2-3 seconds.) But the message > certainly pointed in the wrong direction. Is there more appropriate > language that might indicate that it could also be an effective time-out > because the transfer took too long? Maybe "failed while receiving > responses: clocks are unsynchronized or maximum transfer time exceeded"? > -- > R. Kevin Oberman, Network Engineer > Energy Sciences Network (ESnet) > Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) > E-mail: ober...@es.netPhone: +1 510 486-8634 > Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751 > ___ > bind-users mailing list > bind-users@lists.isc.org > https://lists.isc.org/mailman/listinfo/bind-users -- Mark Andrews, ISC 1 Seymour St., Dundas Valley, NSW 2117, Australia PHONE: +61 2 9871 4742 INTERNET: ma...@isc.org ___ bind-users mailing list bind-users@lists.isc.org https://lists.isc.org/mailman/listinfo/bind-users
Re: Unusual TSIG problem
> From: Mark Andrews > Date: Thu, 09 Dec 2010 09:07:53 +1100 > > > In message <20101208214221.566771c...@ptavv.es.net>, "Kevin Oberman" writes: > > I just ran into an odd issue with a TSIG signed zone transfer. > > > > On occasion I was logging a clocks are unsynchronized message doing a > > transfer from a customer server at a site about 30 ms away. I dropped a > > note to the manager there asking that he look at the his system for a > > time issue. He checked and found no problems. > > > > Today I looked at the problem more closely. I realized that the problem > > was NOT a clock sync issue. They were probably within a millisecond of > > one another. I found the following in the log: > > Dec 8 06:26:18 ns1 named[67170]: zone XX.gov/IN: notify from > > 123.234.1.1 > > #33372: refresh in progress, refresh check queued > > Dec 8 06:31:18 ns1 named[67170]: transfer of 'XX.gov/IN' from > > 123.234.1. > > 1#53: failed while receiving responses: clocks are unsynchronized > > Dec 8 06:31:18 ns1 named[67170]: transfer of 'XX.gov/IN' from > > 123.234.1. > > 1#53: Transfer completed: 1 messages, 397 records, 59674 bytes, 898.462 > > secs > > (66 bytes/sec) > > > > The transfer, probably due to a hardware problem was taking over 5 > > minutes to transfer the zone and RFC2845 suggests tha the difference > > between clocks should be limited to 300 seconds (5 minutes). This really > > means that, should the transfer take over 5 minutes, you get the > > unsynced clocks error. (4.5.2. TIME check and error handling) > > 59674*8/397 = 1202 b/s that's slower than almost all dialup lines. > If this happens regularly use transfer-format multiple-messages which > results in smaller messages and more signatures. I agree that this should not happen. The hardware is obviously broken. It's just that the message resulted in a great deal of wasted effort looking at clock issues when the problem was completely unrelated to that. I'm just wondering if it is worthwhile to mention this possibility in the log message. The more detailed message would have caused me to note the time stamps on the messages immediately and realize what was going on. -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: ober...@es.net Phone: +1 510 486-8634 Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751 ___ bind-users mailing list bind-users@lists.isc.org https://lists.isc.org/mailman/listinfo/bind-users