SolO IRFU quicklooks crash: Value must be a 1x2 vector...

ErikPGJ commented 2 years ago

This issue is not fully researched yet, but I want it documented anyway. It likely needs to be resolved by either ~~@konradSteinvall~~ @JordiBoldu or @ErikPGJ.

Step 1: Latest code?

[ ] This issue is present in the most recent code (i.e. after running git pull this is still a problem).

This issue was found in 6cd2caf which is technically not the latest version, but the relevant code has not been updated since then.

Step 2: Describe your environment

irfu-matlab branch (e.g. master, devel): devel
Matlab version used (e.g. Matlab R2020b): R2020b
Operating system (e.g. Windows 10, Mac OS X 10.15.6, Linux Ubuntu 20.04): Ubuntu 18.04.6 LTS (brain/spis)
[X] A 64-bit installation of Matlab was used.

Step 3: Describe the problem

Expected behavior

Generate Solar Orbiter IRFU quicklooks for the relevant time interval using @konradSteinvall 's code under +solo/quicklook_main.m.

Actual behavior

Code crashes. Two observed cases so far that seem to be caused by the same problem. spis: /home/erjo/logs/so_irfu_quicklooks.2022-05-19_11.29.51.log spis: /home/erjo/logs/so_irfu_quicklooks.2022-05-24_19.08.42.log Ex:

purging solo_L3_rpw-bia-efield-10-seconds_20210209_V01.cdf
purging solo_L3_rpw-bia-density-10-seconds_20210209_V01.cdf
purging solo_L2_mag-rtn-normal_20210209_V02.cdf
  [warning: get_variable(60)] dataobj:get_variable:cyclic_depCyclic dependecy: Epoch <-> Half_interval
  [warning: get_variable(60)] dataobj:get_variable:cyclic_depCyclic dependecy: Epoch <-> Half_interval
  [warning: get_variable(60)] dataobj:get_variable:cyclic_depCyclic dependecy: Epoch <-> Half_interval
  [warning: get_variable(60)] dataobj:get_variable:cyclic_depCyclic dependecy: Epoch <-> Half_interval
  [warning: get_variable(60)] dataobj:get_variable:cyclic_depCyclic dependecy: Epoch <-> Half_interval
  [warning: get_variable(60)] dataobj:get_variable:cyclic_depCyclic dependecy: Epoch <-> Half_interval
exception.message=Value must be a 1x2 vector of numeric type in which the second element is greater than the first element or is Inf.
quicklooks_24_6_2_h, row 44
quicklook_main, row 221
[?1l>STDERR: Command exited with non-zero status 99

Steps to reproduce:

Have not yet tried to reproduce, but the code crashed when executing below calls on brain/spis (see logs):

solo.quicklook_main('/home/erjo/so_irfu-matlab_qli//mission/solar_orbiter/+solo/irf_logo.png', '/data/solo/data_yuri/', '/data/solo/EJ_temp/qli//so_irfu_quicklooks_cron_2022-05-24_19.08.42', '1', '1', '2021-12-20T00:00:00', '2022-05-24T00:00:00');
solo.quicklook_main('/home/erjo/so_irfu-matlab_qli//mission/solar_orbiter/+solo/irf_logo.png', '/data/solo/data_yuri/', '/data/solo/EJ_temp/qli//so_irfu_quicklooks_cron_2022-05-19_11.29.51', '1', '1', '2021-01-24T00:00:00', '2022-01-07T00:00:00');

NOTE: Calls took 20h and 100h, but the duration should be unrelated to the bug.

ErikPGJ commented 2 years ago

Judging from the completed plots, the error occurs when trying to plot data for:

2021-08-29
2022-01-12

thomas-nilsson-irfu commented 2 years ago

I just came across this issue, the error message is typically generated by trying to set the X or Y limits to the same value, one quick example:

fig = figure;
plot(1:2, 2:3, '*-');
ax = fig.Children;
ax.YLim % just output the automatic value
ax.YLim = [min(1), max(1)]; % try to set an incorrect range, result in the error message "Value must be a 1x2 vector of...."

It appears the code in question does have something similar (which could fail if for these dates B has only one value). https://github.com/irfu/irfu-matlab/blob/c54ad3d8c876368bc43ca7a40f79b45978c2e2d9/mission/solar_orbiter/%2Bsolo/quicklooks_24_6_2_h.m#L44

Secondly, it appears the code in question is using hard coded values for fundamental constants, which should be avoided as much as possible.. We have a specific function for that called irf_units (see for instance e99c9c2a3b1d1f8d4cec30d783757f970a0ba839). Things like the following should be avoided whenever possible: https://github.com/irfu/irfu-matlab/blob/c54ad3d8c876368bc43ca7a40f79b45978c2e2d9/mission/solar_orbiter/%2Bsolo/quicklooks_24_6_2_h.m#L17-L18 especially since it seems to suggest proton mass is measured in fractions of kilometres ;)

ErikPGJ commented 1 year ago

Likely other sighting of same bug: Crashes for the call solo.quicklook_main('', '/data/solo/data_yuri/', '/home/erjo/temp/qli_cron/so_qli_cronhelper_cron_MTEST.2022-08-05_13.03.09/interm_output/so_qli_cronhelper_cron_2022-08-05_13.03.09', '1', '1', '2022-01-01T00:00:00', '2022-02-01T00:00:00'); . 64.8 h wall time, but likely unrelated to bug.

Last completed plots for 2022-01-11 which implies that crash probably occurs for 2022-01-12 data (as seen before).

ErikPGJ commented 1 year ago

9c835bd6 JordiBoldu (2022-08-02 12:29:54 +0200) Updated quicklook plots fixes 2022-01-12 so probably fixes bug.

ErikPGJ commented 1 year ago

This appears solved by 7a6b2352 JordiBoldu (2022-08-08 13:41:25 +0200).

irfu / irfu-matlab