NIRSPEC Server Troubleshooting

On occasion NIRSPEC control is inactive due to a server crash. For the user the symptom is that the server window (labeled DO NOT KILL...) stops updating, and a message to the effect

Error - protocol error:  : module [lops.c], function [LOPS_GetRequest]
 -> 1st readlink failed = [-2]

NEW! First, from waimea try running the script ``restart_server''. If this does not work, follow the steps given below.

1. Log into host "waimea" as user "nirspec". Type "psgrep nirspec_server_bin", which will list any server processes that are running; there are normally 3. If there are none listed, type "runserver".

2. If this fails (with a "failed to open TSP link" message), then type "stop_nirspec". Repeat the psgrep command. If one server process remains, then the SCSI-fiber extender box connected to the host computer "waimea" must be power cycled. Afterwards, type "runserver".

3. If this still fails, kill any latent caRepeater processes with the command "kill_caRepeater". Afterwards, type "stop_nirspec", followed by "runserver". If there is still a problem, repeat step 2 one time.

4. If still unsuccessful, check or have someone check the SCSI/fiber boxes at NIRSPEC and in the K2 computer room. The "Link Active" lights should be on at both ends. If not, there is a fiber connection problem.

5. If the link appears good, then reboot the computer waimea.

   ** If waimea is ever rebooted, cycle power on the SCSI-fiber extender box attached to it.**

   Type "runserver" once more.

6. Only as a last resort should NIRSPEC be power cycled. Check the NIRSPEC home page entry under "Service, Power Procedures" for direction.

7. Finally, waimea can be rebooted again, but it should be done from the console, and the SCSI/fiber connection checked with the command "probe-scsi-all" at a breakpoint in the boot script. See Jon Chock or Al Conrad for details on this procedure.
8. Post a short e-mail message to tbida and aconrad with the results of this troubleshooting.