Hi, all -
We recently upgraded our Atlas instance from 2.5 to 2.8 (we’ve been a little slow), and are seeing some odd intermittent errors that I’m hoping form a pattern someone more experienced can diagnose.
Our setup: WebAPI 2.7.6 running on a linux server, connecting to a Postgres database with the CDM data sources and results schema. Atlas 2.8.0 served by a separate application server, which is a back-end proxy for an authentication appliance in the DMZ. Users authenticate to the appliance and are passed through to the application server. For now, we are using simple database authentication within Atlas, to keep the number of interconnections across systems down.
Our problems:
- We get intermittent “Application initialization failed” errors. These don’t generally follow long delays; the client returns with the error within a second or so. Refreshing in the browser somewhere between 1 and 5 times seems to fix the problem. I can’t get a sense of when these are more or less likely to occur. They may appear at the start of a session or after some work has been done.
- Users working on cohort construction get a read-only screen, perhaps as though they were authenticated as someone else. Here’s a screenshot:
- After a user creates a new cohort or concept set, but while they are still working on it, when they try to save, they get an error dialog that reports that there was an issue with saving the concept set or cohort. Again, this appears within a second or two; it’s not waiting a while and appearing to time out. Once it appears, any attempt to save fails with the same error, and the content isn’t in fact saved. Refreshing doesn’t help but the user can close the browser and restart, redo the work, and (sometimes
) save it.
We don’t see any errors logged on the WebAPI side, and can’t spot any distinguishing error messages looking at the browser console display.
Any advice or experience is welcome. Thanks!
We should figure out how to fix it, really - will raise an issue around that