Date
June 24, 2014 12:00-2:00 PDT
< previous (2014-06-10) | next (2014-07-01) >
Attendees
Goals
Discussion Items
Item | Who | Notes |
---|
Design Deep Dive | KTL | -
DM-33
-
Getting issue details...
STATUS
- Level 1 Calibrated Exposure Processing
- Some initial discussion about calibration frames; modified page accordingly.
- Will go to a two-week cycle for these: first week draft is prepared, brief intro, level of detail check; second week is detailed discussion of TBDs.
- Each page should have a list of inputs (and where they come from) and outputs, as appropriate.
|
Storage | Don | - Need to have structure for data and process to move/archive/delete data.
- LSST has 190TB allocated in condo now, 180TB used.
- Could borrow from DES for next month until we can figure out what to purge.
- Mario will arrange a meeting at All Hands to determine what to purge.
- At end of each development cycle, decide what to keep/purge.
|
Compute | Don | - Need to get new hardware with VM provisioning (possibly using OpenStack).
- Also getting experience with containers for launching on borrowed systems.
- Don needs room to have experimental services to gain experience.
- Ticketing system for admins is good as long as it is transparent; should it be LSST's or NCSA's? Would be nice to have comments that are only visible to admins. RT may be the future NCSA system, which can have that feature.
|
Data Center Design | Don | - Don will set up a meeting with Gregory, K-T, and the in-house NCSA experts to see if they can contribute to the design discussions at the All Hands meeting.
|
RFC Process | KTL | - Current process is not broken but needs to be written down.
- System Architect should be able to pull RFCs into SAT (or even veto) as needed.
|
Code Review Process | KTL | - Instead of trying to have everyone read everyone else's code, it may make sense to have categories like "science/apps code", "database code", etc.
- But it's still desirable to make sure that all code can be read by the "average developer".
- It may make sense to have reviews done by someone more experienced than (or at least equally as experienced as) the developer; less-experienced developers may not challenge more-experienced. But then it will be necessary to streamline the review process to lighten the burden on the more-experienced developers.
|
Action Items
- Mario Juric will arrange a meeting at All Hands to determine what data to purge.
- Donald Petravick will set up a meeting with Gregory, K-T, and the in-house NCSA experts to see if they can contribute to the design discussions at the All Hands meeting.
- Kian-Tat Lim will document the RFC process.