Well that wasn’t great
Planned downtime for electric work at our US Data Faciliy on Monday turned into an extended outage when a number of critical systems failed upon cold start. RSP services rely on data products stored and served by those systems (as described here), so as a result DP1 and DP0.2 catalogs were unavailable until Tuesday mid-day (Pacific), DP0.3 were unavailable until Tuesday evening, and DP0.2 images unfortunately are still unavailable at this time.
We are very sorry for any impact it may have had on your work and understand your frustration.
Having some services operable and some not understandably caused some confusion. We debated ahead of the planned downtime whether to keep the platform up even with many data services unavailable or whether to close down access until all services could operate normally. We theorized! We argued! We scratched our heads! But now we can just ask you:
-
I liked the option of accessing the RSP even with many data products unavailable. Keep it up next time. -
It was confusing to have the RSP open for business when many things didn’t work. Lock it down next time. -
Don’t know / don’t care / wasn’t affected / all options are terrible
0 voters
Vote to help us make decisions in the future.
WebDAV service now in beta
Onto better news: as part of this week’s Patch Thursday roll-out, a WebDAV service is now available to data.lsst.cloud users. The RSP will start you your own server on demand. This allows you to mount your RSP home space to your laptop - to copy files, use your favourite editor locally, etc. - with your system’s file browser or a WebDAV-aware client such as Cyberduck (with a name like that, how can you resist?).
We’re still kicking the tires on this one, but check out the docs and give it a try. We’ll poll you for feedback in a future post and feel free to post under the support category if something doesn’t work.
We love all our columns equally but…
(whisper) some more than others.
(ahem)
Our tables are wide. Like, wiiiiiiide. For example the dp1.Object table has 1,296 columns. This is certainly visually overwhelming.
As of yesterday’s Patch, the portal now “pre-ticks” columns that our Data Engineering team suspects you may be most interested in. So if you do not change anything by default your portal searches will return a more compact table.
This saves both mouse strain in horizontal scolling and strain on the database. You can of course always tick additional columns (or even hit the “omni-tick” checkpoint at the top if you really need the full glory).
The same goes for programmatic TAP queries: think before you hit the SELECT * heavy hammer. Everybody will get their results faster if column selection is more intentional.
Never Not Yet On Sunday
The Notebook service’s scratch space on /deleted-sundays is supposed to be purged on, you guessed it, Sundays. We’ve held that off because the first DP1 launch week was short due to the (US) holiday weekend, and because the data availability interruptions messed you around this week.
Staring Sunday July 20th, the filesystem will do what it says on the tin.
User-inspired improvements
Following a question on Support, we realised we should have a list of all the API endpoints in easy reach so you don’t have to trawl the documentation for them. If you hit the API link on the landing page navigation bar, you will find all released data service endpoints in one handy list.
– @frossie for the Rubin Data Services crew