Session destruction relative to loadbalancer setting?

[Russell Beall]

Right.  In this case they are just routing to one server.  Once that was working they were planning on testing with more, but we haven't been able to get the one working.

They sent me the native log for both modes, and surprisingly enough, it looks the same for both cases.  The one thing that stands out is that it appears to be mapping requests to "default" instead of the relevant applicationId.  However, for some reason, the requests are still indeed using the correct settings from the correct ApplicationOverride blocks.

Below is a copy of the native.log for the Active Cookie mode.  The other mode looks exactly the same except for having more lines with the "mapped" message.  The "mapped" message never includes more than the scheme:host/ even though subpaths are being used in the requests.

Regards,
Russ.


2013-01-04 11:38:04 INFO Shibboleth.Config : Shibboleth SP Version 2.4.3
2013-01-04 11:38:04 INFO Shibboleth.Config : Library versions: log4shib 1.0.5, Xerces-C 3.1.1, XMLTooling-C 1.4.2, Shibboleth 1.4.3
2013-01-04 11:38:04 INFO Shibboleth.Config : building ListenerService of type TCPListener...
2013-01-04 11:38:04 INFO Shibboleth.Config : building SessionCache of type StorageService...
2013-01-04 11:38:04 INFO Shibboleth.Config : building RequestMapper of type Native...
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : no resource uri/path/name supplied, will load inline configuration
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : loading inline configuration...
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : added Path mapping (commencementbk)
2013-01-04 11:38:04 INFO Shibboleth.SessionCache : cleanup thread started...run every 900 secs; timeout after 900 secs
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : added Path mapping (myarr)
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : added Path mapping (petition)
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : added Path mapping (myusc)
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : added Path mapping (dms)
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : Added <Host> mapping for http://esdlbtest.esd.usc.edu
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : Added <Host> mapping for http://esdlbtest.esd.usc.edu:80
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : Added <Host> mapping for https://esdlbtest.esd.usc.edu
2013-01-04 11:38:04 DEBUG Shibboleth.RequestMapper : Added <Host> mapping for https://esdlbtest.esd.usc.edu:443
2013-01-04 11:38:04 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:38:12 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:38:21 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:38:30 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:38:39 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:38:48 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:38:57 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:39:06 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:39:15 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:39:24 DEBUG Shibboleth.ISAPI [4364] isapi_shib: mapped https://esdlbtest.esd.usc.edu/ to default
2013-01-04 11:39:30 INFO Shibboleth.Config : shibboleth 2.4.3 library shutting down
2013-01-04 11:39:30 INFO Shibboleth.Config : reload thread finished
2013-01-04 11:39:30 INFO Shibboleth.SessionCache : cleanup thread exiting
2013-01-04 11:39:30 INFO XMLTooling.Config : xmltooling 1.4.2 library shutdown complete
2013-01-04 11:39:30 INFO Shibboleth.Config : shibboleth 2.4.3 library shutdown complete


On Jan 3, 2013, at 4:45 PM, "Cantor, Scott" <cantor.2 at osu.edu> wrote:

> On 1/3/13 6:55 PM, "Russell Beall" <beall at usc.edu> wrote:
> 
>> Has anyone had trouble with SP sessions being destroyed based on what
>> type of loadbalancer persistence is used?
> 
> Yes, but when it was investigated here, the cause was that for whatever
> reason the load balancer cookie didn't take, and it was flipping servers.
> That's not the cause here.
> 
>> I'm trying to help a department where their SP works fine when Source IP
>> persistence is used, but when they turn on "Active Cookie" persistence,
>> the SP calls for the destruction of the session and then tries to
>> generate a new one (resulting in looping).
> 
> When shibd.log is confusing like that, that's when something's in
> native.log that completes the picture. The reason it's removing the
> session should be there.
> 
> -- Scott
> 
> 
> --
> To unsubscribe from this list send an email to users-unsubscribe at shibboleth.net