Hi, I am trying to set up a watchlist for Milky Way globular clusters, however, when I hit ‘Rerun watchlist’, I get ’ 504 Gateway Time-out’ error. I investigated the issue a little bit - it seems like it works all right with a very few number of objects in the list (~5), but as I add more sources, the error starts to appear. At the same time, I have another watchlist with a large number of objects that works all right. The issue may be because of the crossmatch time due to the the large association radii - the problematic watchlist has radii of hundreds arcsecs, and the watchlist that works all right has much smaller radii. But it’s just my hypothesis.
Can you please have a look at it? The Watchlist ids for investigation are 2306 (full list of sources) and 2336 (small sample).
Hi Alex
Let me do a few more tests. But my first thought is that globular clusters are all different sizes, and you would do better to specify a radius for each one. A single maximum radius will find a lot of false positives. So your file would have lines like RA | Dec | ID | radius. This individual radius overrides the single radius for the whole watchlist.
Another way to do it would be a watchmap (MOC file) instead of a watchlist, as we discussed before.
Can I do a few experiments? I will use your file here, is that right? Is ang_size_mass_3 the radius in arcsec?
Roy
Hi Roy,
well, yes, I used custom sizes for each source, I just meant that they are all on a scale 100-1000 arcsec, not something small.
ang_size_mass_3 is in arcsec, yes. Feel free to play around with it, of course.
Hi Alex
Thank you for making the watchlist of globular clusters. I hope it will be OK if we make this one of the “community resources” in the new LSST version of Lasair.
You are right that the “gateway timeout” is because the crossmatch job is too long for the webserver. We try to stop this happening, but your watchlist with its variety of large radii is the first of its kind. Also, your watchlis has exposed some bugs in Lasair, so we will get that fixed, thank you.
However, I have run the crossmatch job from the command line, and get the following results:
2306 – 167 cones, 25750 hits
2336 – 10 cones, 794 hits
Roy
Hi Roy,
the GC watchlist is going to be public, sure. I kept it private for now to do the testing, to not confuse people.
Can you please ping me when you finish troubleshooting so that I could finish creating this watchlist?
I think the watchlist capability is working properly now. Unfortunately there may still be gateway timeout for either a large number of cones, or for large cones, such as the globular clusters. I have added some documentation here and here about the gateway timeout, to please just write to us for these cases.
If there are any further problems or questions, please contact us on this thread.
I would like to copy your watchlist into the “community resources” of the new Rubin version of Lasair, would it be OK for me to do that?
I am particularly interested in what kind of transients you are looking for in globular clusters: the Lasair team would be most grateful if you could write a few words for the Lasair Blog about the nature of your research, could you do that?
Hi @RShA ! I just wanted to check in whether you tested again after this last update, and if you’re now able to proceed with this solution? Please let us know so we can mark as resolved, which may help others navigate similar issues.
Hi @roy, @ChristinaWilliams, sorry, I was a bit busy last week.
So, I checked the watchlist now, and it seems there is still some unobvious behaviour. Namely, when I hit the ‘Rerun watchlist’, I get a message that 25k objects were found - but in the table below only one row is displayed. See the screenshot. I assume it is a new version of the Gateway error, but considering that there is no obvious way to go back to the previous list of matched objects, it seems risky to put on a large watchlist with successfully matched previous observations - anyone can hit ‘Rerun Watchlist’ and the information is lost…
Re: putting the watchlist to the Rubin version of Lasair - absolutely, feel free to do that. I planned to do it myself after testing it on ZTF, but if you prefer to do it, that’s totally fine).
Re: writing a post for the Lasair Blog - sure thing. Give me a few days to draft it. Do you want to have a look at it before I put it in the blog?