| No | Date | Project Code | Observer | Support Scientist | Description & Scheduling Blocks | Response |
|---|
| 1 | 03 Dec 2005 | AGBT05C_031_23 | Kepley | RonMaddalena | The DCR FITS file contained incorrect data for two channels. When the test input was turned on the files contained data similar to the prior scan (2005_12_03_10:49:16.fits). Once the test tone was turned off, the DCR began to operate properly (2005_12_03_10:58:07.fits). (JoeBrandt) | The problem is currently under investigation. (JoeBrandt) |
| 2 | 09 Dec 2005 | T_340_071205 | Minter | Minter | Measurements was hung in Activating after issuing a slew command in astrid and then hitting stop in astrid.
We could only find Nicole who said to restart Measurements and that someone else would look at things later. | The log provides no clues, i.e., everything looked normal and no changes have been made the code. The only recent change has been the host for the databases, but this is the only possible problem we have had. Previously intermittent connection losses to the database have occurred and been fixed by a reset, but nothing like this. Can only watch to see if it repeats itself. MarkClark |
| 3 | 12 Dec 2005 | ABB184 | Braatz | Minter | The Antenna Manager state says there is an error. However, you don't see any error in the CLEO Message screen for the (real) antenna. | This is a known issue with the message system. Workarounds include turning the manager off, then back on, (with axes disabled). JoeBrandt |
| 4 | 14 Dec 2005 | AGBT05C_056_06 | Freire | Balser | When the Antenna Manager reboots the traditional pointing model coefficients are set to zero but the model name is not
set to some default name. So the pointing model was zero even though the model name was Model4c. Of course we measured large pointing offsets but convinced ourselves this was due to ice on the surface until we took a closer look later in the day after other problems that were not related.
In the short term Joe Brandt plans to include a message whenever the coefficients are all zero. In the long term a better solution should be found. | The antenna now includes a message when all coefficients of the pointing model are zero. This should signal to the operator that the pointing model must be re/selected. (JoeBrandt) |
| 5 | 16 Dec 2005 | AGBT_6A_044 | Darling | Balser | The observer was not able to validate a scheduling block for PF2 because the PF2 receiver was not yet installed on the telescope and thus not included in the cabling file.
I understand that this feature was a choice to guarantee the user is connected
to the real system but we may want to discuss alternatives. Maybe during the observing issues meetings. I put it down
here to make sure it does not slip though
the cracks.
| This definitely belongs in the observing issue meeting and the CCB. "Validation" in terms of the config tool if path setup is performed using the /home/gbt or the home/sim environment, depending on whether the user has sourced /home/gbt/gbt.bash or /home/sim/gbt.bash. In either case validating that the config tool can route the IF paths required when the user runs the script in NO WAY insures that on the day of her observing the ifpaths will be 1. available 2. the same ones found during the validation process. This is due to the fact that the cabling file must be used in routing the paths and it can change at any time. I do use several versions of the cabling file for my nightly unit tests,. I must do this in order to be able to run unit tests that are meaningful because 1. if I were to use the "standard" cabling file, I would not be able to insure that the configurations would be consistent from day to day. 2. I would not be able to unit test several backends and receivers; PF1/PF2 are swapped regularly, KA/Q band are removed/installed often, user backends that the config tool must support like LLRadar etc are almost always missing from the cabling file. I have a mechanism in the config tool that allows me to choose what "pickled" cabling file to use. This functionality is not available to the user through astid. I can not endorse giving the user access to the methods from within astrid for a couple of reasons. The first is that the validation process would be completely bogus since the files I use are never synced with the current cabling file. Some of them are 1-2 years old. The second and most important reason I would strongly recommend against this approach is that the user would need to add a special function call in their scripts. If the user forgot to remove the function call and ran the "test" script during actual observation time, the ifpath configuration would most likely be completely invalid. The user would only discover this when either trying to balance or perhaps during the first autopeak. I think this approach is a recipe for disaster. I think we need to consider what it really means to "validate" a script. Is it syntatic validation or more? We can validate the script syntactically, and in most cases we can do "more"; providing the user with a "warm" feeling that her telescope configuration is within the "normal" operating limitiations of the telscope. But since the cabling file and the device health file can change without notice, I don't think we can ever absolutely gaurantee ahead of the observation time that any given configuration script will work. (Unless we get rid of the cabling file AND solder everything together  ) |
| 6 | 20 Dec 2005 | HF commissioning | Mason | Minter | hi all- last night I ran into a problem I never suceeded in working around, which was that the archivist kept inhibitating an activate b/c of an illegal value. As a result I removed the archivist from the scan coordinator-- a slight problem as it's useful to have co-registered weather (e.g. wind) info to help make sense of or flag your high frequency gbt data. It wasn't sufficiently urgent to wake up Toney but, I feel like this is a known problem with a known solution-- what should I do next time this happens?
Brian
| Call someone. A quick check of the message log shows that a message was asserted stating the manager name for one of the samplers was blank. (JoeBrandt) |