TAP service down?

Support
#1

Dears, is the TAP service down? This small query:

from lsst.rsp import get_tap_service

service = get_tap_service("tap")
service.search(
    f"SELECT * FROM dp1.CcdVisit LIMIT 100"
)

Fails with DALQueryError: Query timeout exceeded for synchronous execution. The job has been aborted. Please use /async endpoint for long-running queries.. Async queries on the other hand keep going without finishing the job. Any idea when we will get online again? Thank you.

#2

Hello, I had a quick look and found a stuck process in the Qserv database server. Service should now be restored . Sorry for the inconvenience!

2 Likes
#3

Thank you Fritz.
While the simple query above is indeed working now, this query still is not:

from lsst.rsp import get_tap_service

service = get_tap_service("tap")
job = service.submit_job(
    "SELECT * "
    "FROM dp1.CcdVisit "
    "WHERE CONTAINS(POINT('ICRS', ra, dec), CIRCLE('ICRS', 53.160, -28.100, 1.000))=1 "
    "AND band IN ('g', 'r', 'i') "
    "ORDER BY obsStartMJD"
)
job.run()
job.wait(phases=["COMPLETED", "ERROR"])

I used it over and over in the past and usually it took just a few seconds.

Side note, I tried testing the simpler query with an async call and it’s finishing just fine.

#4

Hi Giuseppe,

I’ve had another look at this – I can see the associated translated queries executing successfully in our backend Qserv database, but there appears to be some issue in our TAP service layered above or in the plumbing in between… Interestingly, I can get the query to execute successfully all the way through for a smaller search radius (although the query radius as you have written it is certainly reasonable and should be working!)

I’ll flag some folks working with the higher-level services to take a look, but unfortunately support is somewhat thin today do to the national US holiday.

1 Like
#5

This is more for @fritzm and our RSP developer team trying to diagnose the issue than advice for you, @deppep, because using SELECT * shouldn’t be a problem on the CcdVisit table, but – I did some testing and found that the same query statement but with specific columns selected (as below) runs as an async query in about 3 seconds, but swap in SELECT * and it’s still executing after 6m.

query =  """SELECT ra, dec, band, expTime, visitId, obsStartMJD FROM dp1.CcdVisit 
         WHERE CONTAINS(POINT('ICRS', ra, dec),
         CIRCLE('ICRS', 53.160, -28.100, 1.00)) = 1 
         AND band IN ('g', 'r', 'i') ORDER BY obsStartMJD"""
1 Like