Bug #13105
closed
Missing federation config makes the multi-site search's session manager fail to log in
Added by Lucas Di Pentima almost 7 years ago.
Updated over 6 years ago.
Release relationship:
Auto
Description
It seems that if API's remote_hosts configuration is not declared on the config file, the discovery document doesn't get a useful default value (In this case, would be {}) and also the Javascript code from SessionDB assumes that remoteHosts will always be defined.
We could fix this problem on both sides, adding proper error messages to the Javascript code if it happens to be that a newer workbench is working against an older API server.
- Target version changed from 2018-02-28 Sprint to 2018-03-14 Sprint
- Target version changed from 2018-03-14 Sprint to 2018-03-28 Sprint
- Target version changed from 2018-03-28 Sprint to 2018-04-11 Sprint
Seems like this shouldn't be an error -- Workbench should behave as if config is empty/false if those remote_* entries aren't in the discovery doc. This should only happen with an old API server -- API server config already defaults to {}.
- Status changed from New to In Progress
Instead of having 4 variations of this, could we fill in both missing values once in discoveryDoc(), and fix the two cases that don't call discoveryDoc() so they do? I expect this will also eliminate some redundant requests by giving discoveryDoc()'s cache more opportunities.
-m.request(baseURL+'discovery/v1/apis/arvados/v1/rest').then(function(dd) {
+db.discoveryDoc({baseURL: baseURL}).map(function(dd) {
...
})
suggest
-db.discoveryDoc({baseURL: session.baseURL}).map...
+db.discoveryDoc(session).map...
the rest LGTM, thanks!
- Status changed from In Progress to Resolved
- % Done changed from 0 to 100
- Related to Bug #13851: Workbench multi-site search bug added
Also available in: Atom
PDF
Updated by 
Updated by Tom Morris 
Updated by 
Updated by