Overloaded shibd questions

Nate Klingenstein ndk at signet.id
Thu May 20 21:06:27 UTC 2021


> There are warnings in the authorization and exporting steps when it can't locate a session because of an exception raised (e.g. a ListenerException because shibd won't answer in time).

So, to my dismay, we aren't seeing any WARNs or ERRORs in native.log(other than that we need to add the secure flag to cookies and the occasional IP address mismatch, etc., normal stuff).  We are seeing spontaneous, occasionally serial re-initialization of the ISAPI filter(chosen over the module by the deployers) several times per day, but at times that aren't contemporaneous with the looping.  native.logger has been customized, but not in such a way that I think it would squelch those messages.

We're also dealing with a large enough set of attributes that using the SessionCache as an interim measure could lead to excessively large cookies if we persisted all the attributes the application can use.

I think that rather than trying to track this down, just implementing stickiness and not using a single shared shibd instance might be more expedient, since that's the long term fix anyway.


More information about the users mailing list