Reports 1-1 of 1 Clear search Modify search
DGS (General)
takahiro.yamamoto - 12:06 Tuesday 17 November 2026 (37862) Print this report
HWP device is unreachable
I heard guardian was stopped during the operation of INITIAL_ALIGNMENT guardian and then found that HWP device was unreachable from k1hwp.
Guardian log is as follows.
2026-11-17_02:52:21.237913Z -oStrictHostKeyChecking=no -i /kagra/apps/agilis/id_ed25519_agilis_p_callback x.x.x.x 1 1 ******* IFO_REFL
2026-11-17_02:52:21.652699Z VendorID 4173
2026-11-17_02:52:21.652699Z ProductID 12294
2026-11-17_02:52:21.655325Z Traceback (most recent call last):
2026-11-17_02:52:21.655325Z File "/kagra/apps/agilis/agilis-p_control_with_callback.py", line 387, in
2026-11-17_02:52:21.655325Z node, channel, axis, cur_pos = main(device)
2026-11-17_02:52:21.655325Z File "/kagra/apps/agilis/agilis-p_control_with_callback.py", line 298, in main
2026-11-17_02:52:21.655325Z device.init(iserial,axis)
2026-11-17_02:52:21.655325Z File "/kagra/apps/agilis/agilis-p_control_with_callback.py", line 46, in init
2026-11-17_02:52:21.655325Z raise ValueError('Device is not connected')
2026-11-17_02:52:21.655325Z ValueError: Device is not connected


I also tried to directly access the USB devices on k1hwp, but valid device cannot be found. This means that we need to try rebooting k1hwp once. After the reboot, finding 0 position of HWP is necessary and it requires to stop INITIAL_ALIGNMENT guardian once. But current INITIAL_ALIGNMENT guardian has user code error as follows.
File "/opt/rtcds/userapps/release/lsc/k1/guardian/INITIAL_ALIGNMENT.py", line 4, in
from lsclib import check_io, check_als_x, check_als_y, check_ir_x, check_ir_y, check_ir_x_or_y, check_prmi3f, check_src1f, check_omc, check_srmi1f, check_drmi
ImportError: cannot import name 'check_ir_x_or_y' from 'lsclib' (/opt/rtcds/userapps/release/asc/k1/guardian/lsclib.py)

Because of this error, some guardian command doesn't work and I cannot understand why INITIAL_ALIGNMENT guardian works now with such fundamental code bug.

Unfortunately we have no time to fix it today's daytime due to the fire drill. It will be fix later.
Comments to this report:
takahiro.yamamoto - 17:55 Tuesday 17 November 2026 (37865) Print this report
[Ikeda, YamaT]

We found finally that the HWP issue was caused by missing packages on which the script depended, so we installed the necessary packages on k1grd1. Installation manual of the guardian server was also updated (klog#17316).

Since testing the Guardian code does not extend to checking external code called via `subprocess` (as this requires running the code on an actual device to be considered a valid test), this issue had been overlooked. Because it increases the difficulty of migration, the use of `subprocess` in Guardian should be avoided. Alternative methods include `Threading` and `concurrent.future`.

By the way, we found two lsclib.py as
- lsc/common/guardian/lsclib.py
- asc/k1/guardian/lsclib.py
Which one is called depends on the Python path order: if a current hard-coded configuration file is used, the `lsc` version is called, while in an automatically generated environment, the `asc` version (which appears first in alphabetical order) is called. Which code was actually used in O3GK and O4a/c? Science Mode may be no longer reliable. Stop such sloppy quality management and operation.

hirose.chiaki - 18:07 Tuesday 17 November 2026 (37866) Print this report

As a temporary workaround, I renamed `lsclib.py` in the `asc` directory to `lsclib_asc.py`.

Search Help
×

Warning

×