2/25/03 La Silla * Lunch *Introduction to La Silla Observaotry by Mr.Percy Glaves and Mr. Jorge Ibsen 100 staff total with 2 shifts, at the same time, 60 staff is on duty Admin, Lodging people are hired in La Serena Tech, Support Scientists, Engineer hired in Santiago or Internationally 22 to 24 staff are woking in weekend *Observation mode Service observation is carrried out by Support Astronomers 20 to 25 % of NTT telescope time is in service mode. Other is in visitor mode. 100% of 2.2m telescope time is in service mode. Historically, remote observation was test at CAT (Coude Auxiliary telescope). Now CAT is closed and no remote observation is made now. NTT had also capability of remote observation, but after upgrade of NTT software to VLT version, no capability for remote observation. Therefore, no remote observation is made here. Service observation is made by queue programs provided by Garching. Garching made queue arrangement and La Silla runs observation based on it. Final decision on service observation is made by a staff astronomer in La Silla, depending on the sky and facility conditions. *Software staff composition Total: 12 to 20 staff Software engineers 6 (Had he more budget, like to hire 8 to 10) Sys admin 3 Database admin 2 Team leader (sys admin) 1 Others are 12 Mechanics La Silla: 70% for development and commissioning 30% for maintenance cf. 100% maintenace at Paranal Specialists from Garching help for commissioning and development. Some development for VLT software is made in La Silla. Most La Silla instrument software is developed here. Instrument software and TCS for VLT was tested in NTT in 2001/2. *Instrument development Development of instrument software takes two years. But last instrument (TIMMI2) software was development in six months, thanks to CCS (Common Control Software). In average, software for one to two instruments is on going. Instrument software developed by an outside contractor is 'bad idea' because it causes a lot of problems unless sending a local expert to work with an outside contractor from early stage to sure compliance with CCS and UML (Unified Modeling Language) - ESO requirement model - . "Waterfall Model (Marunage)" causes bad results. He recommends prototyping and continuous testing to avoid bad surprises. GUI buider, Panelator, is used under Tcl/Tk to make GUI quickly only in La Silla (& Paranal). Observation control windows is on X-Terminals. Garching makes two releases of CCS(Common Control Software). One for prerelease and the other for upgrade; La Silla uses 2nd upgrade version. They use X terminals with 1GB Ethernet adapters to put machines in the next room. *Telescope Info Seeing size on the night was .6 to .7 arcsec for 3.6m telescope. Alluminization is taken place every two years. Dome flat screen is a mere cinema screen. 14 telescopes including a radio telescope in La Silla. out of them, 3.6m, 3.5m(NTT), 2.2m, 1.5m, 1.5m, 15m radio telescope are working. Control room for NTT and 2.2m in the newly built facility was seen. In coming 2 months, 3.6m telescope control are moved to the new control building. They don't use XML. ALMA uses it. VLT is 'a nightmare' because of C++. ACE(Adaptive Communication Environment) and TAO(The ACE Objects) ACE may replace RTAP in future? CCS is used at La Silla and ALMA. *NGAS NGAS hardware is in 2.2m dome. Removable 80GB hard drives are used. Next version will come with 200GB. NGAS can write on hard drive faster than CD-R or DVD. Development of NGAS software was quickly (2 months) because they used Python, XML and stand protocol of HTTP. Python is so efficient. *Interview of Gaetano Andreoni (Head of Engineering department) on ISO 9001 Started reengieering six years ago. Reorganization - SciOpe, Eng, Infrastructure, Software teams with goal of ISO 9001 certification Reporting system, remedy, documentation project. One part-time internal auditor per department. ESO hired external consultant with $30000. Aim at overall excellence. Proctivity of La Silla is higher than La Paranal even though budget is much lower. Jacconi, former director, said "I found the recepe productivity doubles when I cut budget by one third." Evaluation is done every year for astronomers and every four years for engineers. ESO considers clients not to be astronomers but to be the community. Index of success is measured not by the number of papers but by the citation number for astronomers. Vision of La Silla is the best observatory in the world and he cited as an example kitchen budget cutting by hiring outside contractors and La Silla has the best kitchen. After retirement of the chef, he worries about getting another person of skill. He hires local people because they are highly motivated. Even if La Silla maintains NTT only and shuts down all other telescopes, number of staff needed will not change significantly if it wants to keep the level of service. Gift instrument which had been written with no standard needs one staff astronomer to operate. He stressed standardization and software compliance. They cannot have ad-hoc software. La Silla will not build any instrument incompliant. Emergency Response Team 1 engineer: on call 1 technician: on call 1 drau(hydraulic) mechanic: until midnight 1 IR/CCD detector technician: on call 1 system admin/application programmer: until 3 a.m. 2% down time, 1/4 of problems requires physical intervention. Eng staff in La Silla 2 database 3 sys admin (2 unix, 1 windows) 1 leader 12 mech (2 engineers, precision mech, 2 air LN2, drau mech) 6 Instruments (2 cryo, 2 detectors, 2 optics) Total budget for ESO is 100M euro 7% of ESO budget is for La Silla operation cost (6M euro) 1/3 of La Silla budget is to maintain infrastructure. * Dinner *3.6m telescope Commissioning of HARPS HARPS(High-Accuracy Radial-velocity Planetary Search) is developed by a Swiss group. Budget is provided by the team, they are given 100 nights yearly for 5 years. Normally ESO pay HW, and consortium pay for man power in return with GT. User manual must be completed by the end of commissioning. Technical/maintenance manual can be completed later. Document may be extracted from source code with full comments. Learned P2PP, BOB and layout of the control room. 14 displays 1 Log 2 TCS 1 AG 1 SkyCat for guiding 1 BOB/HARPS 1 SkyCat for real-time browser from HARPS 1 P2PP 4 for instrument/observers (Top shelf above Log and TCS) 1 weather monitor 1 dome tv monitor Slit positioning takes 2 to 2.5 minutes. Several iteration is applied but readout time is smaller thanks to the smaller readout region. It takes 90 sec to read out an entire 2kx2k pixels. Quality control first stage: on site by staff astronomer or observers second stage(offline): at Garching. After checking, data is copied to DVD or CD-R and sent to the observer. * NTT/2.2m control room Made in Nov. 2002. After the commissioning of HARPS, the control room of 3.6m will moved to the same buidling. Very quiet. One staff astronomer of NTT confessed that the gathered operation room is not favored because of too many people. Many people don't like fluorecent bulbs. SPIKE is not used because 1) Users can select program manually 2) NTT software lost compatibility after it is changed to comply with VLT software. In service mode, support astronomer manually confirms the completion of the set of data for targets and calibration. If no target is found, this target must be skipped and reported to Garching, where someone contacts PI. *Visual observation of the Southern Sky and LMC/SMC. 2/26/03 * NTT Could not see the telescope, water sealant was applied. * 3.6m telescope Interview by Emilio Barrios (Deputy team leader of Sci.Ope.) Received four documents 13 telescopes in La Silla 3 telescope teams (3.6m/NTT/small telescopes): 1995-2002/6 Members Astronomers: little more than number of instruments Night Operators 1 electronics/electric engineer 1 electronics/electric technician In 2002, changed to a single team model. Small telescope were closed and moved their software to ESO software and hardware standard. Engineering, mechanical and software staff moved to the engineering team which covers all the telescopes. 3.6m telescope has four instruments. One at telescope, 3 others at Lab. Maintenance (calibration, upgrade)is made while the instrument is offline. One TIO(Telescope/Instrument Operator) operators both the telescope and instrument. Staff astronomer works from noon to midnight. *Instrument exchange After the instrument change, one night is spent for system check. Instrument exchange takes place at least once every month. Top ring is exchanged, one night for set-up is assigned to pointing check, etc. To change an instrument, the instrument is moved to downstairs to the observing floor and left there for 24 hours to cool down. *staff Report is entered to remedy and read next morning by another staff. Staff astronomers performs calibration data acquisition for the previous night. Operators perform reloading instruments and test everyday. Performs weekly test for maintenance (detector, etc) using "detector test gui and data logging system" The team built the software for test specifically. Enginnering time is scheduled every six months (for one week). Daily meeting is held after lunch. If no problem, no meeting. Weekly meeting is held every Wednesday afternoon to make an official report of any problem which happened the past week. There is one to two hour overlap with the members of the next shift. Night shift starts 12 degree before sunset and ends 12 degree after sunrise. Day crew shift starts from 8 to ? Operators are all Chilean. Legal work hour per week is 48 hours. Night shift is one week per month. 20% higher salary for night work. *Training First one week is for learning the control panels and the command/data flow The next week is for TCS (shift with noon till midnight) If no problem, they can operate the telescope in a month. Trainee works in midday-midnight shift from the second week and learn alpha-delta and so on. Learn instrument operation while it is in the Lab. May be able operate telescope/instrument alone in six months but most can do that earlier (3 to 6 month). Selftraining follows after that. Remedy can be read from internal WWW. *Mr.Jorge introduced the software Lab. 14:00- VxWork simulator running on VME board computer with 68k or PowerPC. ESO is moving to real-time Linux. Software structure ------------------------------ Application (TCS,ICB,DCS, etc) ------------------------------ CCS ------------------------------ gnutool | VxWorks | RTAP ------------------------------ Directory structure is fixed. $VLTROOT - CCS $INTROOT - can be changed by mgr Common library(CCS) is sit in $VLTROOT. CCS update twice a year. Source code size is unknown to Mr. Jorge. Source code control is currently done with CMM at Garching, but is to be moved to CVS at Garching, which will be used for ALMA because CVS is more popular than CMM. Because CVS is so popular that common domain is publicly available. ALMA replaced RTAP(HP specific real-time OS) to ACS (ALMA Common Software; CORBA, TclTK, Perl, Eclipse, etc). CSS exports source and library as a package. They develop source code and test code, and create document simultaneously. *Development of software 1 User requirement phase uses "UseCase Analysis". 2 Preliminary design phase (PDR) employs prototyping and sometimes, Use Case Analysis, and Documentation starts. 3 Final design phase (PDR) creates final source, user manual and maintenance manual. 4 Commissioning is system delivery phase and pre-acceptance test is run. 5 Final acceptance test. Simulator is used in phase 2,3,4 and 5 RTAP has some compatibility with HP-UX ver.12 but it does not matter because ESO is leaving RTAP. *Training for new person Training period is six months. But currently new hire can produce source code in three months. Software person receive training at least once a year. Jorge goes to Garching once a year. ESO has 60 to 70 software people in Garching with 100% development duty. La Silla has six software staff but preferably hopes to have 8 to 10 because one person has more than one project and because it is ineffective. Six programmers have two shifts of three people in group in every week. ESO extended the source of softwear people pool. Six staff is bare minimum. Six programmers should be for maintenance. Outside developers must use a standardized requirement process (Use Case). ALMA has heavy users/experts of Rational Rose and UML. DFS(Data Flow Software) is an outside CCS which contains P2PP. BOB is Broker of Object Block which resides in CCS. *Two documents were given. http://www.eso.org/Projects/VLT/sw-dev http://www.eso.org:8082/Development/Computing * Data handling (Data handling administrator: Anton Schemrl) Data Handling diagram was provided. DVD with raw data is sent to Garching and is archived in DVD format in La Silla (more than two year ago, CD-ROM used.) Observation data are handed to observer from Garching. Data transfer using rcp(rhost) for easy connection inside observatory. Sybase is replicated in La Silla and Garching with possilbly some complicated issues which has not been completely solved. Data is released to public after one year. Open data is available through VO. Everybody is moving to the local pipeline. Completion of Dataset which is in use at Subaru is made Offline not online. It is still in developping phase. Original: J. Noumaru Revised: T.Sasaki