-
Notifications
You must be signed in to change notification settings - Fork 50
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Unreliable DOI resolution for some DOIs #947
Comments
If it is taking over a minute to resolve requests to doi.org I suspect that their service was down or degraded. Some of those links are now working and respond fast, for example: curl -L -H 'Accept: application/x-bibtex' https://doi.org/10.5281/zenodo.8184298 We are caching the requests to DOI (as of a few weeks ago), so these are faster for local builds, but that is not persistent across CI builds. We don't currently retry in the build, but could add that logic. doi.org is the canonical way to access this information, so I don't think we should change that. Have you consistently come across this, or was this a point in time DOI issue with their service being down? Probably the best way forward is the retry+backoff and raise a more sensible error if links are taking more than a few seconds to resolve (this should be on the order of milliseconds...). |
I don't yet have the data to know if I've consistently come across this across days, only across hours. So maybe we can use this issue to see if others chime in with "me too" or not. Maybe an easy thing would be something like:
|
There are multiple improvements that we have made to the DOI resolution:
This means we are trying doi.org at least twice, with two different content-negotiation strategies, caching successful results, and have updated the error messages (and internal ways we are saving citation data, now in CSL not bibtex strings). I am going to close this specific issue, feel free to reopen if these DOIs remain troublesome. 🤞 |
Description
It seems like DOI resolution is occasionally flaky and returns different results depending on the run. I noticed that locally my build would sometimes hang on building a page with many DOIs, sometimes not. In the built page, a subset of DOIs would have their metadata added as a reference, and others wouldn't be.
Not sure how to reproduce this reliably, but as an example here are two builds from a recent experiment I was running at using the MyST CLI (from this czi report repository).
Here's one run that builds the repository with the action logs linked below:
https://github.com/2i2c-org/report-czi-2021/actions/runs/8118256669/job/22192166630#step:6:19
And a relevant excerpt:
And from another run with more-or-less the same content, we get a different subset of DOI resolution errors:
link to run
Proposed solution
Perhaps this could be fixed with either:
The text was updated successfully, but these errors were encountered: