Reports 1-1 of 1 Clear search Modify search
AdV-COM (automation)
bersanetti, carbognani - 1:50 Wednesday 06 January 2021 (50366) Print this report
Automation shift: system-wide PySat installation, PythonVirgoTools upgrade, reactivation of the *ARM_LOCK Metatron nodes, minor works on INJ_MAIN

At the beginning of the shift we made:

  • site installation of PythonVirgoTools v5r0p2 (for both anaconda python2 and python3)
  • installation in /virgoApp and run from there of PySat v2r1

Made quick check for automation restarting all SUS nodes and driving them to ALIGNED or MISALIGNED; note that PySat is still logging verbosely, to verify that all is working as expected; still, the logs are outside the /virgoLog/ area (in /data/dev/logs/VirgoOnlineTest/SatServer/) to avoid any issue about disk space. After further checks, it will be reverted to standard logging in the standard place.

We made also some tests steering with the AlignementUI GUI confirming good behavior but also the presence of the Cm reply processing errors, further investigations are needed but this is confirmed as a minor problem.

In order to be compliant with PythonVirgoTools v5r0p2 the statement:

sender = pvt_utils.native_str(sender)

needed to be added into all Cm handlers of python based servers to solve "returned non-string (type bytes)" error

For this reasons the following updated version of packages were also installed, made default in VPM config and corresponding servers stopped/restarted:

  • PySb v4r2
  • PySbEth v1r2p1
  • PyLnfs100 v2r1p1
  • PyAgilis v3r1p1
  • PyGenie v2r1p1
  • PyArtic v2r0p1

For some of the servers (ex: SWEB_MotSwitch) , non impacting commands were sent in order to double check correct commands management.

During the whole test session no occurrences of the tango timeout error were reported into the PySat server logging. Few occurrences of NaN response from GetGain() could be spotted but those were managed by the automation nodes without problems.

Then we worked on the reactivation of some Metatron nodes:

  • we reactivated the NARM_LOCK and WARM_LOCK nodes and managed to robustly lock the arm cavities with them; a few notes:
    • the engagement of the lock was smooth; also the boost filters and the drift control were tested and worked reliably, especially the latter as we are now using the new PySat server;
    • we updated the configuration files with the last used parameters from the ISC shift of Mon 4th;
    • we did not update, upgrade or polish the code of the nodes, as at first we want to check that everything is in order before starting any development; any future development will need consideration about backward compatibility with the O3 configurations;
    • from now on the nodes SUS_NI, SUS_NE, SUS_BS, SUS_WI and SUS_WE are managed again by the appropriate arm node; this means that the corresponding suspensions will be automatically steered as needed;
    • SUS_PR and SUS_SR are not managed and have to be checked manually; in particular, the aligned/misaligned positions of SUS_SR are to be considered as work in progress and not usable without a preliminary check;
    • as this feature was managed by upper, still un-commissioned nodes, bringing the arm nodes to DOWN will NOT bring the general safety switch LSC_ENABLE to zero, so it has to be done manually via iPython/VPM; this will avoid unlocking both nodes at the same time when only one is asked to, and it is not a safety issue as we don't need it with these very basic configurations: the safety switches of the single loops in the DOWN state should be enough, but it is best to document it anyway;
    • the "single bounce from NI" configuration is achievable by using the two nodes and selecting the states MISALIGNED_WEST_ARM and MISALIGNED_NORTH_END in the appropriate node.
  • then we did some minor work on the INJ_MAIN node:
    • we checked the first part of the FmodErr procedure manually, namely the check of the mistuning itself, and the actuation on the LNFS;
    • we changed the way we change the modulation frequency via the LNFS: since now we have cm handlers for that, we replaced all the calls to external scripts with the corresponding cm command; the underlying code of the two is essentially the same, so this change is transparent;
    • the logic of both the check and the tuning via the LNFS are working, while the calibration of the latter obviously changed over time and will need to be updated; since we can steer the LNFS only up to a point, this will probably be done in coincidence with the future work on the mechanical tuning, which has been in any case hardcodedly bypassed for the time being;
    • INJ_MAIN is still not managed by any node, and it has to be checked manually before using it or other nodes. IMC_RESTORED is the target state, as usual when the node is moved manually.
Comments to this report:
bersanetti - 11:03 Thursday 07 January 2021 (50377) Print this report

The note about the LSC_ENABLE flag also applies to ASC_ENABLE, as it was also managed to upper nodes. It has to be = 1 for any alignment to work (drift control included) and it is not set to zero automatically but, again, the loops are sufficiently safe on their own. While there is no need for LSC_ENABLE, ASC_ENABLE needs instead to be put back to 1 manually in case of process restart,

Search Help
×

Warning

×