Displaying report 1-1 of 1.
Reports until 12:03, Wednesday 06 May 2015
H1 CDS
hugh.radkins@LIGO.ORG - posted 12:03, Wednesday 06 May 2015 - last comment - 12:51, Wednesday 06 May 2015(18275)
Keeping SDF Green--CHANS NOT FOUND list

SDF Cleanup

If channels are removed from your model/code whatever, unless you take your system down to the safe state and makeSafeBackup to generate from scratch a new safe.snap (and most of us know the problem doing that causes,) the SDF system will go yellow for CHANS NOT FOUND.  Okay, maybe no big deal but green is better than yellow when scanning our configuration control tools and we should green these up at earliest opportunity.

There is a way to clear these out with having to take the system down to a 'safe' state for the new snap.

You can just go to the SDF TABLE and while there you might as well look at the CHANS NOT FOUND list and make sure they make sense (yes, we did remove those channels from the model, e.g.)

Then from that medm, open the SDF SAVE SCREEN

Insure the SAVE TABLE selection is TABLE TO FILE and the FILE OPTIONS SELECTION is OVERWRITE

Press SAVE FILE

This step overwrites the safe.snap file removing the not found channel records.  Little wierd though, on the SDF_RESTORE screen (available from the TABLE screen as well,) it does not show a Modified file detected message, but, it has updated the time, as if it had 'Restored' to that time.  But you will notice that your list of Not Found Channels have not cleared.  Maybe this is explicitly intentional or is maybe a trivial bug.

Now on the SDF_RESTORE screen, press LOAD TABLE, and, all the NOT FOUND channels should clear.

However, there may be remaining NOT FOUNDs that will crop back up:

If the NOT FOUND CHANNELs list contains fields as opposed to just records, that is H1:HPI-ETMX_GUARD_CADENCE.HIGH (field) versus H1:HPI-ETMX_GUARD_CADENCE (record), this process will not remove the field lines from the safe.snap.

Still need to test this (Barker is on it) but, maybe the next time the front end starts, these fields will again become a CHANS NOT FOUND and our green lite will revert to yellow.

This happens because the FE code reads the safe.snap and strips out the field lines and saves them in a safe_alarms.snap file in the target area.  Then when it saves the safe.snap, this field list is just tacked onto the end of the safe.snap record list.  So your safe.snap will always contain these non existent channel field lines.

The fix to this problem is the following, before doing the steps above, (you could did it after but you'd be saving & loading twice) go to the target area and remove the appropriate field lines from the safe_alarm.snap file (currently writable only by controls.)  Once these are gone, proceed as above in the SDF SAVE SCREEN.

Comments related to this report
jameson.rollins@LIGO.ORG - 12:13, Wednesday 06 May 2015 (18276)

For what it's worth, all of those old "...GUARD_CADENCE" channels should be completely purged from all models.  That is old deprecated guardian infrastructure.  I think it has all been purged at LLO, so maybe LHO just needs to update the affected models.

jeffrey.kissel@LIGO.ORG - 12:51, Wednesday 06 May 2015 (18278)GRD
@Jamie -- indeed -- we've updated the hepitemplate.mdl which has these old GUARD channels remove and compiled / installed the BSC HEPI models this past Tuesday (LHO aLOG 18223). So, Hugh is just documenting how to get rid of them the last known hold out -- the safe.snaps.

The integration issues that tracks the removal of these vestigial organs are here:
SEI II 922
SUS II 921

They've been marked closed for some reason -- I think it's because again and again we've thought we'd gotten rid of them all, but continue to find them lurking about.
Displaying report 1-1 of 1.