|Ticket Number:||20081119_1||Ticket State:||CLOSED|
|Ticket Opened:||2008-11-19 17:38||Ticket Closed:||2009-01-07 13:16|
|Ticket Description:||Problems on the DWDMlight link Lausanne - Manno|
Problem Description:Since approx. 1 pm on Nov 19, 2008, we observe periodic input errors on our interface swiEL2 T1/4 (10GE from Manno).
Our investigations reveal an additional 15dB of attenuation between the active nodes at Lausanne/EPFL and Martigny RERO. Due to the characteristics of our optical amplifiers, this large attenuation gets compensated in the chain of amplifiers and can hardly be noticed at CSCS (at a distance of 5 amplifiers). However, as the minimum S/N ratio is exceeded in one node, we see sporadic crc errors on the 10GE channel whereas the parallel 1GE channel is still working properly.
No change in attenuation can be observed on the CWDM link swiEL2 - swiMY2 which uses the second fiber of our pair.
No direct impact on any site is expected except a minor performance degradation for high bandwidth streams between Manno and Lausanne. Redundancy is still guaranteed everywhere.
Affected:From 2008-11-19 13:00 until undefined
Impact: no impact
All connections using the affected fiber are running error-free since 20 hours. Ticket closed.
The fault has been located at the SBB station in Martigny. A patch cable was not correctly inserted. Yesterday, new labels were attached to the patch panels. We assume that the patch was never correctly inserted and that it has been shifted during the labelling work. The patch has been correctly inserted at 14:20 and the optical power values are OK now.
A technician from SBB is expected to arrive in Martigny at about 14:00. Meanwhile we proved that the fiber under suspicion between SBB Station and PoP Sunrise is rented by SWITCH from SBB (though the cable is in the property of Sunrise).
From the results of our OTDR measures we expect a fiber problem at the Martigny station. We try to get access to the network room at the station. SBB netop is informed.
All traffic has been rerouted to prevent the use of this link.
To allow fiber measurements - the fiber needs to be interrupted.
Characteristics of bti amplifiers better understood. It's now pretty sure we have a problem with the fibers which needs to be localized.
RERO in Martigny contacted. They did not do any work near our rack today.
Connectis contacted and asked them to check sanity of our bti amplifiers in Lausanne and Martigny. Chris Watts confirms that no hardware problems seem to exist. Optical input powers are 15 dB too low on both nodes.
First investigations start. Optical input power problems detected on amplifiers in Lausanne and Martigny. Surprisingly, the power drop at 1pm can be observed only on transponders in the west (Lausanne, Geneva), not in the east (Manno)!
For all questions about this ticket, please send mail to email@example.com
or call +41 44 268 15 30.