Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] BioPortal: Questions

Jennifer Leigh Vendetti vendetti at stanford.edu
Thu Oct 18 18:27:41 PDT 2018


Hi Lieven & Simon,

Since this issue is proving somewhat difficult for us to nail down / fix, I’ve started an issue in our GitHub repository where you can follow our progress:

https://github.com/ncbo/bioportal-project/issues/100


On Oct 18, 2018, at 12:36 AM, redcap <redcap at UZGENT.be<mailto:redcap at UZGENT.be>> wrote:

Hi Jennifer,

  *   We're still leaning toward some kind of a local front-end caching issue.

As a backend developer I can confirm that a 500 HTTP error is always a backend issue.


Yes, I understand that. This is why I asked Simon in a previous correspondence to remove the BioPortal front-end from the equation altogether, and issue a call directly to our REST endpoint for the Event class in ATC. Simon reported that the call was successful, which I initially took as an indicator that the problem was originating from something else in the stack.

The request for Simon to access the Event class in an incognito window was essentially to allow us to be totally certain that the error page wasn’t cached.

Thanks for trying that out and letting us know the error still occurs.


The error message can always be found in the server logs. If an incorrect argument is given a 4xx should be thrown. Not a 500 which means internal server error.


I didn’t offer an update on the log files yesterday because I hadn’t yet finished my examination of them.

Our REST endpoint currently handles about 20 million requests per week. The UI traffic is load balanced across two servers and the REST API traffic is load balanced across three servers. We lack the resources to have any sort of log analysis software put into place. This means that tracking down particular errors is a manual process that requires us to sift through large files on multiple machines, and can take some time. In the end, I did locate a stack trace in one of our UI logs that allowed us to make some progress in understanding why the exception is occurring. Details were recorded in the GitHub issue I referenced above.


I can reproduce the problem as well, so it is not browser specific. Steps to reproduce are:

  1.  Login to bioportal. (This problem does not happen if you’re not logged in).
  2.  Browse to the https://bioportal.bioontology.org/ontologies/ATC/?p=classes&conceptid=http%3A%2F%2Fpurl.bioontology.org%2Fontology%2FSTY%2FT051&jump_to_nav=true page.
  3.  Expected: The class shows
  4.  Actual: An error page is shown:

<image001.png>


Thanks for reiterating the set of steps. We did figure out today what we’d missed, which is that the error is only occurring for a particular type of account (non-administrative). Also, the error isn’t specific to ATC and the Event class, but rather is reproducible when searching for any of the UMLS Semantic Type classes while viewing any UMLS ontology. I recorded an updated set of steps to reproduce in the GitHub issue.

Kind regards,
Jennifer

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20181019/3157a636/attachment.html>


More information about the bioontology-support mailing list