Downloads Not Working

a durkee

2 years ago
updated 2 years ago
Not a bug

I've been using OSM Boundaries to download Geojson files without issue for a few months now, up to and including last night. However, as of this morning around 10 AM PDT, all of my downloads are failing. The website keeps indicating that it is thinking, until eventually it gives up and shows an error. It does not matter what I am attempting to download- even just trying to download an outline of Andorra will fail.

In Firefox, the error is:
"The page isn’t redirecting properly An error occurred during a connection to osm-boundaries.com. This problem can sometimes be caused by disabling or refusing to accept cookies."

The error in Chrome is similar.

I have tried changing the settings of the downloads, changing which data to download, changing the date of the OSM database, and wiping my cookies and reaccepting all cookies again. I have tried using the direct download link, the button, and even downloaded curl, where it continued attempting to download for 10 minutes before I closed it.

Could someone please confirm that it is not just me, and I am not losing my mind?

If it is just me, what can I do to fix it? If not, should I wait for an update? (and will it be announced somewhere?)

Thank you!

Pinned replies

Magnus

2 years ago
Answer
Not a bug

It is working, but the workers are busy.

https://osm-boundaries.com/Status

There are jobs created that requires quite a lot of time, which isn't normal. But it works as intended. We will however look into optimizing some jobs.

About 70 jobs have completed the last 10 hours.

Reply Link

Replies 4

Newest first
Newest first Oldest first

2 years ago

I just wanted to update this thread. A new release has been made today that significantly decreases the time to generate the files that was blocking the queue earlier. Hopefully we shouldn't see any jobs that requires more than an hour now, with the exception of the jobs importing new databases. With the old code some jobs could take weeks, we expect that a job that earlier took a week, should not run in less than 15 minutes. A decent improvement. :)

There are ten workers. Worker 1 is reserved for jobs with higher priority (lower number). All user triggered jobs has such priority.

It's too hard to predict if jobs are quick or not. The real fix is to optimize what's blocking the queue, but due to lack of man power it won't happen this month. Its very likely that you won't notice any issues in 1-2 days from now.

Am I correct in reading the info on the Status page you linked above- that all 9 workers have been busy for the last 16 hours, during which time no one has been able to submit a job?

Would it be possible to request that 1 of the 9 workers eventually be modified to only allow quick jobs? I can't be the only person who submits small jobs, and right now 7 or fewer users are blocking literally every request (workers 3 & 9 and 6 & 8 were booked simultaneously, so they're probably the same users)