KAGRA Logbook

DetChar (General)

hirotaka.yuzurihara - 17:11 Friday 15 August 2025 (34837)

Temporary suspension of uploading segment files from Kamioka to DQSEGDB

[YamaT, Yuzurihara]
As reported in klog34882 and klog34808, during 8/13, some of the frame data were missed in the stored directory, which the segment script will read. As a result, the produced segment files are partially incorrect. YamaT-san recognized this issue and shared the information quickly. So, I suspended uploading the segment files to DQSEGDB before uploading the segment files of 8/13 including the incorrect information. Now, the segment files by 8/12 were uploaded to the DQSEGDB and they should be correct. (We already shared this situation with the administrator of DQSEGDB.)

I checked the segment files of 8/13 located in the Kamioka computer and found that the time when the Frame Writers hung was removed from the science mode segment. (The GPS or UTC times are shown in klog34822.) However, this is an incorrect situation.
I attached the plots of the IFO guardian, which was read via k1nds0 and k1nds1. Figures 1~3 show that the interferometer was actually in the OBSERVING mode at that time. So, it's necessary to regenerate the segment file of 8/13.

The current script to produce the segment files can read past data via k1nds2 and can produce segment files. Therefore, I plan once the gaps in the missing data are filled in the Kashiwa cluster, I will regenerate the segment files and restart the upload to DQSEGDB.

The segment files at the Kashiwa cluster require the action to reproduce. Note that the time of the missing frame files in the Kashiwa cluster is different from that in Kamioka.

Images attached to this report

34837_1755243426_Screenshot from 2025-08-15 16-34-07.png

34837_1755243429_Screenshot from 2025-08-15 16-34-50.png

34837_1755243433_Screenshot from 2025-08-15 16-35-24.png

Comments to this report:

hirotaka.yuzurihara - 15:37 Friday 07 November 2025 (35527)

This is a report of last week. Sorry for late.

As reported in klog34823, the gap of the frame data in Kashiwa cluster (called system-B) was filled on 10/10. So, I regenerated the segment files of 8/13 data, via k1det2.
The working directory is /users/yuzu/work/20251030_segment@k1det1. I used the conda environment of igwn-py38, which is same environment as the regular segment production.
I asked the final review of the generated segment files to YamaT-san. After his confirmation, I will copy the files to the /users/DET/Segments and ask the upload of segment files to DQSEGDB.

NOTE

To reproduce the segment files reading 1 day data, it took around 30 minutes.
After the trouble of the missing frame data, I just suspended the copy process from the temporal directory to /users/DET/Segments. So, the other segment files (after 8/13) are regularly generated and stored in /users/DET/tools/Segments/Script/Partial.
From August 15 (in UTC timezone), the interferometer was unfortunately not in the OBSERVATION state due to the laser issue (klog34839 and klog34840). So, we don't need to care the segment files after August 15.

hirotaka.yuzurihara - 22:56 Thursday 13 November 2025 (35573)

[YamaT, Yuzurihara]

We have completed the regeneration of the problematic segments and the final content check. The segment files have been copied to the production directory. The regular upload of segments to DQSEGDB will be resumed from tomorrow.

Details

For the period 8/14–11/12, segment files had continued to be created by crontab; these were copied from the intermediate directory to the production directory, keeping the original timestamps. The used script is /users/yuzu/work/20251030_segment/cp_segment.sh.
Additionally, for 8/13, 9/30, and 10/2, the automatically generated segments had issues. So, we reproduced these segment by readin past data via k1nds2. The regenerated segment files were copied from /users/yuzu/work/20251030_segment to the production directory, keeping the original timestamps. The used script is /users/yuzu/work/20251030_segment/cp_segment_update_08-13.sh.
In particular, the segment of 8/13 had an issue during SCIENCE mode and it requires correction. The others had an issue not during the SCIENCE mode, but we replaced them just in case.
From tomorrow morning, we will resume the crontab that copies newly generated segments to the production directory.