WP8108

Jonathan, Dave:

The DAQ Fujitsu 10gigabit switch was reconfigured today in preparation for a dual-data-concentrator upgrade.

In the old configuration, all ports belonged to VLAN-102, and there were some gaps in the port assignments.

In the new configuration, a new VLAN-110 was created and assigned to the first 6 ports. All old ports were shuffled to the right to clear these first 6 ports, and port gaps were filled. This required a data interruption for all DAQ machines except h1tw1.

We did the upgrade in two phases.

PHASE-I, assign and activate ports 7-16 to VLAN-102, move DAQ machines one-at-a-time to their new port, waiting for each one to come back before proceeding. h1dc0 was last to be moved, since its return caused all the other systems to restart again. From a user's point of view, this looked like two DAQ restarts (with no configuration change). Where necessary, ports 17-24 were assigned to the default VLAN-100 and disabled.

PHASE-II, disconnect the FE-DC0 link from direct Netgear-to-h1dc0 to Netgear-VLAN110-h1dc0. This was done in two steps to ensure h1dc0 continued to run throughout. First the fiber from the last Netgear port (50) was disconnected at h1dc0 and rerouted to port-1 of the Fujitsu. A new 2M fiber from port-3 Fujitsu (VLAN-110) took its place at h1dc0. While this was being routed about half the front end's data was bad. This was most of the corner station. Last step was to disconnect Netgear-port49 to h1dc0 and replace this with the second 2M fiber from Fujitsu-port-4.

Data went good again, no restarts of mx_stream or h1dc0 were needed.