james.batch@LIGO.ORG - posted 15:24, Tuesday 24 November 2015 - last comment - 15:42, Tuesday 24 November 2015(23711)
Modified psinject scripts, put psinject back under control of Monit
Dave, Jim We modified the /etc/init.d/psinject script on h1hwinj1 to ramp the gain on the H1:CAL-PINJX_CW filter from 0 to 1 over a 10 second interval when starting psinject. Similarly, there is a ramp down of the gain from 1 to 0 over a 10 second interval when stopping psinject. Although this does not prevent a transient if psinject dies unexpectedly, it does help when Monit restarts the process. Since the /etc/init.d/psinject takes longer to finish execution, we also changed the /etc/monit.d/psinject file to include longer timeouts. Modified files are included as attachments. We tested the revised setup by starting and stopping the psinject using Monit and /etc/init.d/psinject, and while under control of Monit, we tested killing the psinject process to ensure that Monit would restart it. Since the scripts now set the gain and ramp time of the filter, we also changed the SDF file to no longer monitor the channels to avoid problems from SDF differences. We left the CW injection running.
Non-image files attached to this report
Comments related to this report
I have modified the CDS O1 overview MEDM to make calex a hardware injection system and made the calcs a standard, non-hardware injection system. This means that calex is green when the CW injection is running, and calcs is green when no injections are running.
Guardian will still permit excitations in either calex or calcs and keep the IFO in observation mode. If we are to no longer run any hardware injections to calcs, we may want to remove it from the exception list.