Note: items marked ** are highest priority to fix, since they impact observing
SUMMARY LIST: STATUS
A. Non-software items:
5. Intermittent (and possibly spurious) readings of low unresolved
voltage from Renishaw quality signals from the three
stages, even when wheels are parked. Suspect that
Renishaw quality signals are OK, but that there is
a marginal connection (e.g., incompletely seated pin
in connector) to the analog board and/or Galil
controller that monitor them.
4 Jan. 2000 update: The dewar filter wheel (DWFILREN)
seems to be the worst of the three in terms of these
dropouts. They have gotten so bad that we have turned
off the pop-up alarms in the GUI for this keyword.
**6. Hatch occasionally failed to open or close in a timely fixed?
manner at various times during all 3 engineering runs.
Usually, rotating to a different instrument PA provided
a work-around, but this still causes some lost observing
time.
The procedure for freeing a stuck hatch has been documented.
In December 1999 we found a bad seal on the cylinder. The
cylinder was replaced, and new, more robust cylinders placed
on order. Twenty cycles of the hatch showed no problems.
Air pressure is currently set to ~110 psi.
9. Collimator actuator #1 exhibits somewhat higher backlash unresolved
than the other two collimator actuators. This may bear
some relationship to item #13, or it may be totally
unrelated.
10. Multi-filter trays are extremely difficult to remove once pending
inserted into any of the three wheels. Need to design and
fabricate a fixture to allow easier removal.
**11. Several ESI ion pump controllers have failed due to the monitor
same capacitor going bad. Need to see whether most
recent ion pump controller replacement has solved this
problem.
4 January 2000 update:
So far so good. The current ion pump seems to be working
well.
12. CCD shutter often fails to open sufficiently fast monitor
following long periods of inactivity. Problem appears
to clear once shutter is exercised once or twice.
**13. CCD readout suffers from low-level fixed-pattern readout pending
noise whenever the instrument is positioned (relative to
gravity) so that collimator actuator #1 is carrying most
of the weight of the collimator mirror and thus putting
out relatively high levels of torque. Suspect inductive
coupling between motor controller cable to actuator #1
and the high voltage cable to the ion pump. Need to
re-route the HV cable to ion pump so that it is not in
proximity to any motor control cables. This problem
might also bear some relationship to item #9, or it may
be totally unrelated.
14. Spectral lines are tilted by 1.4 pixels along their length. pending
Need to rotate CCD dewar by corresponding amount so that
spectral lines will be aligned with the row/cols of CCD.
Dewar rotation scheduled prior to February ESI run.
15. Full complement of filter/mask trays has not been pending
completely machined. Full complement of longslit masks
has not been machined. Need better mask material?
16. Snout/baffle needs final alignment plus documentation of pending
alignment procedure if imaging needed for 12/28/99 run.
Otherwise, this item will be resolved as soon as CARA
baffles the tertiary tower, as they have agreed to do.
B. CCD sub-system software items: (Kibrick)
1. watch_ccd: compute SYNOPSIS from SYNOPFMT and re-write pending
SYNOPSIS at end of exposure
2. watch_ccd: if OBJECT="" at exposure end, then in the pending
FITS header set OBJECT= ($TARGNAME) where $TARGNAME
is the current value of the DCS TARGNAME keyword
3. lickserv: test removing the unlink to see if this solves pending
the problem with trying to overwrite a file that you
don't own
4. Fix lickserv so that it never overwrites existing files? pending
5. Suspect that a race condition exists if an ABORT occurs pending
too close to the end of an exposure such that it happens
in the interval between EXPOSIP going false and WCRATE
going true. In that case, the ABORT is seen by watch_ccd
but ignored by rccd, so the header gets deleted but the
image still reads out and is written with partial header.
29 December 1999 update:
May have been fixed with updates to the CCD VME software.
This problem is rare enough that we are not closing it out
yet.
6. The rccd process on the VME crate gets into a confused pending
state if the observer brings up a dashboard and attempts
to read out an image without first having started up
lickserv. If this happens, the CCD VME crate needs to
get rebooted and WCRATE set false to clear the problem.
This problem should be clearly noted in the ESI Engineering
Notes, the User Manual, and the ESI Quick Start.
C. Motion control sub-system software items: (Tucker/Burrous)
2. Can't get to position "5C" on the slit mask wheel. pending
The software limits for this stage may be overly
restrictive and need to be checked.
3. Reconcile lamp keyword names with hardware reality deferred
4. Reconcile temperature keyword names with hardware reality deferred
5. Need to set negative gain for AIRPWPSI air pressure pending
sensor added by CARA to compensate for inverted wiring.
6. Fix the dispatcher so that it doesn't send second-level pending
move completion messages in the case of motions initiated
with the KTL NOWAIT option.
D. ESI GUI items: (Clarke)
11. Need to reconcile conflicting requests from Goodrich, partial fix
Bolte, and Miller regarding setting of CCD window.
Goodrich has requested separate buttons on CCD panel to
select full-frame readout .vs. imaging mode window .vs.
low-dispersion window. Bolte has suggested that WINDOW
be set to an appropriate value whenever ESIMODE is
changed. Miller says to simply do this via setups.
Partial fix is to provide the "Quick Frame" button.
14. Consider re-enabling MOVE button when exposures are consider
paused, so as to allow interactive focus-move
exposures (i.e., exposures where telescope is moved
and focus adjusted within the same exposure).
Need to consider risk of observer's accidentally
initiated a move when not intended.
15. If dashboard is running on the same machine as lickserv, pending
(e.g., kanaha), at start of exposure scan the OUTDIR
directory for files whose names begin with the value of
the OUTFILE keyword and determine the highest frame
number that is currently written in that directory. If
FRAMENO is less than or equal to the value, pop up a
warning box.
18. Provide a new hook to the alarm mechanism that is pending
triggered when a keyword is about to be written with
a new value (but before the write occurs). This would
allows us to check changes to TTIME that result from the
UPDATE CCD button being hit. In this way, if TTIME were
set <= ELAPTIME during an exposure and the UpdateCCD
button was then pressed, we could PAUSE the exposure
and pop-up a confirm box (as we do for the STOP button).
Alternatively, a simpler compromise solution would be
to simply pop-up a warning whenever desired_TTIME was
set to less than ELAPTIME and an exposure was currently
in progress. The pop-up would warn the user that
hitting the UpdateCCD button would result in premature
termination of the exposure currently in progress.
4 January 2000:
The pop-up warning suggested in the second half of the
description above has been implemented.
21. Should we be designing and implementing a GUI set-up consider
editor? If not, need to provide some mechanism for
adding non-standard keywords to a set up. Perhaps
consider having the dashboard pop-up an editor
session (e.g., vi) to allow the user to manually
edit a setup selected from within the GUI?
23. In order to properly reflect the order in which incoming deferred
photons hit the various elements within ESI, the
collimator icon should appear before (i.e., to the left
of) the ESI mode icon.
25. Under conditions of long latency and/or high packet loss pending
between the GUI and the instrument (e.g., when running
the GUI locally in California), the dashboard can lose
contact with the back end (yet not close its traffic
connection). It ends up in a strange state where it
can do no KTL I/O operations, yet it is still alive.
The visible symptom is that the heartbeats stop and
you get error misleading error pop-ups, because the
back end is really still running.
26. Certain bogus hatch conditions can leave the hatch pending
pushbutton stranded outside the hatch rectangle. We
have not been able to reproduce this, and await a more
precise description of the conditions that result in
this problem.
27. We need to come up with a better way of handling the pending
default values that are set whenever the observer
changes spectrograph. Currently, we have values from
namespace (e.g., Clear_I) that are hard-coded into the
GUI logic, but those values can be changed whenever
inconfig is run to reconfigure the contents of the
wheels. If the OA changes these names, then the GUI
code breaks. Perhaps this functionality needs to be
implemented via some alarms logic that is abstracted
into the database. This is really part of a larger
problem of abstracting expert knowledge regarding
the characteristics of the removable elements currently
contained in the wheels.
28. The dial gauges don't have adequate labeling, so the pending
tick marks aren't particularly useful. These are
"canned" widgets we got for free. To make them more
informative we either need to re-write their C code
or adopt a variant improved version from Hobbs.
29. Non-repeatable behavior is observed within the CCD detail pending
grid widget, with the behavior changing from one invocation
of the GUI to another. Sometimes this widget erroneously
highlights fields when desired == actual (and also
propagates this erroneous state to upper level buttons),
sometimes it fails to highlight fields when desired !=
actual, and sometimes it displays unexpected values
(e.g., blank text) in the desired field. Suspect that
non-repeatability may be a function of the timing of
asynchronous events (e.g., X events and/or broadcasts
of ESI keywords) that occur during GUI startup. Similar
problems have been observed with the User Config detail
panel. When the dashboard is run remotely from California,
both problems are likely aggravated by the delays and packet
loss associated with the trans-Pacific cable. Due to the
long time required to start up the GUI when running
remotely, this is a difficult and time consuming problem
to troubleshoot. Instrumenting the GUI to keep track of
nesting depth might shed some light on the problem. The
extent to which this problem is apparent when running the
GUI from Waimea is not currently known, and needs to be
tested.
E. Electrolog for ESI items: (Deich)
2. Fix spurious error messages about DCS keywords pending
F. Global software items: (Allen)
4. Preserve the existing hand-crafted /vx on koki, then pending
replace with a cleanly built /vx checked out from
repository
G. Software test suite / telemetry items : (Clarke)
1. Need to resolve ktl_watch restart strategy pending
2. Possible changes to collimator backlash tests to better pending
characterize the anomalous behavior of collimator 1
3. Ktest suite needs to be revised so that it adjusts its pending
tests of ordinal positions in response to the dynamic
configurations in .../dyna/mapRON.cfg
4. Make ktlwatch logging less verbose so it doesn't eat so pending
much disk space on kanaha
H. Hardware/software integration testing: (Kibrick/Goodrich)
3. Double-check centering of wheel positions pending
I. Software items requiring collaboration/joint solution with CARA
1. Fix the hostnames on the ESI private network. pending
Currently, on the private net, kanaha is still
esinstb.ucolick.org, koki is still esupvb.ucolick.org,
and the VME crate is still esi5cb.ucolick.org. Although
there is no great harm in this (since the private net is not
advertised via DNS), it is likely to confuse everyone in the
future. Requires changing VME crate EEROM to fix this.
(Kibrick / CARA)
2. koki and kanaha can send e-mail but they can't receive it defer
(CARA)
**3. Eliminate cross-mount dependencies so that koki and pending
kanaha don't routinely mount other CARA machines, such
as pelekane. (CARA)
4. OA's and instrument scientists need a simple method pending
(Openwin pulldown?) for tailing relevant log files
(e.g., trafficlog, dispatcher logs, etc.) on koki
but without having to log into koki. Set up some
menu items which ssh the tails from the local machine
at which the OA or IS is sitting.
(Allen / CARA)
6. Need to find a way to make the boot-disk-mirroring script pending
sufficiently robust so that it can be run as a cron job
(Kibrick / CARA)
10. Get Lupton's official version of tkrose installed on pending
kanaha. Tkrose currently installed on kanaha has been
slightly hacked by De to get it to work.
(Clarke / CARA)
11. Installation and configuration of tklogger on kanaha to pending
monitor conditions such as disk full, etc. (Kibrick /
Goodrich/ E. Chock)
12. Fix .openwin-menu files for esi, esiN, and esieng accounts pending
so that esicon is invoked on kanaha via ssh1 rather than
ssh. (CARA)
13. Fix MALIGN scripts so that the don't set PREPIX=0. Rather, pending
they should set KEEPPREP=false. Also make sure MALIGN
scripts don't leave inappropriate values set if they are
aborted. (Goodrich)
14. Need to get a better mechanism for eliminating processes pending
left over from previous users. Need clarification from
CARA re: the goals and purposes of the esiN accounts so
we know how these accounts are intended to interact. This
was an action item to CARA at the 12/8/1999 SCC meeting.
(Allen / CARA)
15. Update ESI standard instrument setups to reflect most done
recent name change for the clear positions (e.g.,
change "Clear_(I)" to "Clear_I", "Clear_(S)" to
"Clear_S".
16. Fix DCS terminal server (tserver4) configuration so that pending
it doesn't inhibit ESI CCD VME crate from rebooting.
(E. Chock)
J. Flexure compensation software: (Kibrick)
**2. User interface for steering spectra via collimator tilts in progress
3. Echellette spectrum re-centering via image centroids deferred
4. Need to implement wrapper script around esicon so that done
it survives restarts of the DCS or ESI software.
K. Figdisp wishlist items: (Allen)
3. Via a command line argument and/or X resource, select an consider
alternate method for defining the "stats box". If enabled,
instead of using left mouse click to mark upper-left and
right-mouse click to mark lower-right, hold down the left
mouse button and do a rubber-banded drag to define the box.
If the new command line argument and/or X resource is not
specified, the existing mouse button behavior would be
preserved.
4. Provide a method for setting the CCD readout window by pending
marking a box on the figdisp bit map graphics window.
Possibly combine this option with item 3 so that the
right mouse button is used to request setting the readout
window (to be confirmed via a pop-up?).
5. Provide a method for displaying the current readout window pending
within the full CCD. (Currently, we do not have any
graphic window which always displays the full CCD, so this
might require yet another window. Alternatively, perhaps
the LOC window could be changed so that it always shows
the full CCD, and then uses two different colored outlines
(e.g., red and green) to indicate the current readout window
versus the current pan/zoom window.
7. Provide a function for mapping figdisp cursor position pending
into an echelle order (if appropriate) and wavelength.
8. Review key mappings for figdisp and reassign then from pending
L?, R?, and other Sun-specific keywboard function keys
to the universally-available and currently unused
alphabet keys.
----------------------------------------
Software Punchlist items carried over from June 16, 1999 punchlist
Status as of Dec 13, 1999:
Note: items marked ** are URGENT to fix, since they impact observing
SUMMARY LIST: STATUS
A. Non-software items:
4. Temperature sensor readings corrupted when motors are run not fixed
B. CCD sub-system software items:
3. Bogus TEMPDET values if traffic and/or infoman down (Kibrick) pending
C. Motion control sub-system software items:
6. Full KTL notification if commanded to current pos'n (Tucker) defer
7. Fix motion hangs due to switch in/out of JOG mode (Tucker) defer
8. Can't reach ordinal position 0 on shuttle stages (Burrous) defer
13. Uncomment AIRALARM test routine in Galil firmware (Burrous) defer
D. ESI GUI items:
20. Clean up codegen error messages (Clarke) cosmetic
E. ESI Log items:
F. Global software items:
G. Software test suite items:
6. Fix tolerance for "near" test within LDFLT limits check test pending
(see e-mail from Kibrick dated: Tue, 25 May 1999 15:13:46
H. Hardware/software integration testing: (Clarke/Burrous/Tucker)
I. Software items requiring collaboration/joint solution with CARA
1. ktl_close fails to tag address as stale (Allen/Chock/Lupton) pending
3. Memory leaks in ktl_dispatch (Allen/Chock/Lupton) pending
------------------------------------------
Software Punchlist items carried over from February 5, 1999 punchlist
Status as of December 15, 1999:
SUMMARY LIST: STATUS
A. Non-software items:
6. Dewar contamination (Stover/Brown/Wei/Kibrick) resolved???
B. CCD sub-system software items:
10. Make XFLIP usage consistent with LRIS (Kibrick/Allen) pending
C. Motion control sub-system software items:
No items carried forward
D. ESI GUI issues
No items carried forward
-------------------------------------------------------------------------------