Add Condor `request disk`

Closes #181

See merge request jameson.rollins/locklost!111

Camilla Compton authored 2 years ago and Jameson Rollins committed 2 years ago

Found that both "analyze" and "search" logs are < 0.1 MB, "online" logs to the lockloss directory so is request_disk needs to be barely anything for that. Spoke to Dan and confirmed a blanket request_disk of 10Mb for all methods (online, analyze, search) is fine. This is much larger than we expect but ensures our jobs won't be rejected. Fix removed the "must specify request_disk" message when tested on pcdev1.

Closes #181

Ads plot added for lower guardian states

Closes #153

See merge request !109

Define variables for issue 179

See merge request !108

Austin Jennings authored 3 years ago and Jameson Rollins committed 3 years ago

closes 179

temporary remove ZM1 ZM2 channels from saturations so test doesn't fail

See merge request !107

Changing websites tag query to be a dropdown with ability to select multiple tags

Closes #120

See merge request !104

Only refine time if new lockloss time is before ISC_LOCK lockloss

Closes #109 and #168

See merge request !105

move history.py plugin to end as takes longest and has no dependances

Closes #175

See merge request jameson.rollins/locklost!106

This delays the update until the actual iteration starts.

simplify to make things a bit cleaner

See merge request jameson.rollins/locklost!100

Yannick Lecoeuche authored 4 years ago and Jameson Rollins committed 4 years ago

This adds an additional plot to the summary page that indicates which suspensions tend to saturate with each other. The 1st suspension to saturate in a lockloss goes on the y-axis, and the 2nd-5th saturating suspensions go on the x-axis. The code only looks at suspensions that saturated before the refined lockloss time and does not allow repeats (that come from UR, UL, LR, LL).

Closes #126

See merge request jameson.rollins/locklost!96

Yannick Lecoeuche authored 4 years ago and Jameson Rollins committed 4 years ago

Fixes the shift check in the summary code by assigning timezones for LHO and LLO. Also cleans up a large portion of code with defaultdicts, for-else statements, and any other improvements I could think to make.

Improve zoomed refine plots

See merge request jameson.rollins/locklost!99

Yannick Lecoeuche authored 4 years ago and Jameson Rollins committed 4 years ago

As mentioned in #161, when a zoomed refined time plot has a threshold crossing outside of the plot window, the channel data outside the window does not get plotted. This is because the code is specifically selecting data that falls within the window. This code plots all the data and chooses xlims based on the refined time and window boundaries.

This generally shouldn't be necessary, as we don't expect good refinement to yield times more than 5 seconds before the unrefined lockloss time, but it's nice to have consistent plots.

Adding check for fast LOCKLOSS_DRMI -> ACQUIRE_DRMI_1F -> LOCKLOSS transitions

Closes #109 and #150

See merge request jameson.rollins/locklost!91

Yannick Lecoeuche authored 4 years ago and Jameson Rollins committed 4 years ago

Add logic to the history plugin to check if ISC_LOCK_STATE went from the lockloss state we're interested in (generally between 103-202) to LOCKLOSS_DRMI (3), then up briefly to ACQUIRE_DRMI_1F (101) before LOCKLOSS (2). In order to create the refined time plot correctly, the history followup moved before the refine followup in execution order. Additionally, I added a setter to change transition_index to the new values (I also removed the assertions, because they would cause an error whenever this condition was met).  This is only for H1.

Closes #150

Search for and tag FSS oscillations before lockloss

Closes #109 and #154

See merge request jameson.rollins/locklost!94

Yannick Lecoeuche authored 4 years ago and Jameson Rollins committed 4 years ago

Look at PSL-FSS_TPD_DC_OUT_DQ for dips below 0.3 within 5 seconds before lockloss and creates tag if it finds any. These FSS-related locklosses are fast, and the glitch is expected to occur very close to the refined lockloss time. For this reason it depends on a strong lockloss refinement, which may preclude LLO from getting good use of this followup.

Check and tag if Observatory mode is Commissioning or Maintenance

Closes #109 and #121

See merge request !95

Split up refine_time into two functions so the calculation can be called separately

Closes #159

See merge request !97

Camilla Compton authored 4 years ago and Jameson Rollins committed 4 years ago

This is mainly a copy/paste to split the refine.py function refine_time into two functions:

* find_transition: Takes an input of the time window and dictionary containing information about the channel needed for refinement (channel, std, min); then outputs the threshold used and refined time.
* refine_time: Does the job of the old refine_time function but calls new find_transition function to do the actual refinement calculation.

The reason for this change is that in the future I would like add a plugin to look at speed or order of light falling off PDs. This plugin could then call the function refine_calculate rather than copy/pasting refinement code.

Only change made to logic is on lines 58-60, where now if the refinement has already been done, the previously refined time will be used to center the search window rather than the transition_gps. This will not make a difference for the normal refinement but should speed up future plugins that use the refinement calculation as the search window could be smaller.

Closes #159