[ 
https://issues.apache.org/jira/browse/DIRKRB-692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16355619#comment-16355619
 ] 

Fabiano Tarlao commented on DIRKRB-692:
---------------------------------------

I cannot provide a final test for your project, I have not understood how your 
integration tests work. I think you spawn a local kdc instance with predefined 
domain users.

I have provided a normal test KrbClientTest.java that assumes a kdc is already 
running somewhere and that class has constants that should be initialized 
properly... 
[|https://issues.apache.org/jira/secure/DeleteAttachment!default.jspa?id=13136579&deleteAttachmentId=12909628&from=issue]it
 also loads the kdc configuration from an available kerb5.conf. You can also 
change that part to a getResource... new File(...getURI)... I have currently 
no-idea of your best practises about. Hope useful

Regards

> 1)sgtTicket clientPrincipal is not initialized + 2)KrbClient fails to store 
> SGT ticket in cache file
> ----------------------------------------------------------------------------------------------------
>
>                 Key: DIRKRB-692
>                 URL: https://issues.apache.org/jira/browse/DIRKRB-692
>             Project: Directory Kerberos
>          Issue Type: Bug
>    Affects Versions: 1.1.0
>         Environment: Linux Mint 17.1 + Netbeans 8.1 with a Maven Project + 
> kerb-core 1.1.0 in Windows AD (KDC is Windows 2016 Server)
>            Reporter: Fabiano Tarlao
>            Priority: Major
>              Labels: easyfix
>         Attachments: KrbClientTest.java, bug1.diff, bug2.diff
>
>
> Two bugs that interact each other
> h1. 1)
> *SgtTicket*, returned by *KrbClient.requestSgt(..)*, has a _null_ 
> *clientPrincipal* field (unassigned). Perhaps this is not mandatory but your 
> code assumes this property is populated (see later). I highly suggest to 
> populate this field.
> I have wrote a workaround that overrides the *requestSgt* method and works 
> for the *USE_TGT* case:
>  
> {code:java}
> @Override
>     public SgtTicket requestSgt(KOptions requestOptions) throws KrbException {
>         SgtTicket requestSgt = super.requestSgt(requestOptions); 
>         TgtTicket tgt = (TgtTicket) 
> requestOptions.getOptionValue(KrbOption.USE_TGT);
>         if(tgt != null){
>             requestSgt.setClientPrincipal(tgt.getClientPrincipal());
>         }
>         return requestSgt;
>     }{code}
>  
> h1. 2)
> Creating a new credential cache file when storing only one SGT (service 
> ticket) fails. (i.e., trying to create a new cache file containing only one 
> SGT and no TGT)
> Invoking *KrbClient.storeTicket(sgtTicket, File)* fails for this reason (here 
> is the original code in *KrbClientBase* class, my comments in RED ):
> {{public void storeTicket(SgtTicket sgtTicket, File ccacheFile) throws 
> KrbException {}}
>  {{        LOG.info("Storing the sgt to the credential cache file.");}}
>  {{        if (!ccacheFile.exists()) {}}
>  {{            createCacheFile(ccacheFile);{color:#ff0000} //Correctly 
> creates a new file but...{color}}}
>  {{        if (ccacheFile.exists() && ccacheFile.canWrite()) {}}
>  {{            CredentialCache cCache = new CredentialCache();}}
>  {{            try {}}
>  {{                cCache.load(ccacheFile);{color:#ff0000} //..this line 
> EXPLODES cause it tries to initialize from an empty file, the unexistent file 
> case is not managed correctly{color}}}
>  {{                cCache.addCredential(new Credential(sgtTicket, 
> sgtTicket.getClientPrincipal()));}}
>  {{                
> cCache.setPrimaryPrincipal(sgtTicket.getClientPrincipal());}}
>  {{                cCache.store(ccacheFile);}}
>  {{            } catch (IOException e) {}}
>  {{                throw new KrbException("Failed to store sgt", e);}}
> {{        } else {}}
>  {{            throw new IllegalArgumentException("Invalid ccache file, "}}
>  {{                    + "not exist or writable: " + 
> ccacheFile.getAbsolutePath());}}
>  {{}}}
> Here follows my proposal/fix, this code correctly manages the MIT ccache file 
> creation for one SGT, please note that this fix assumes that bug 1 is already 
> fixed:
> {{public static void storeTicket(SgtTicket sgtTicket, File ccacheFile) throws 
> KrbException {}}
>  {{        LOG.info("Storing the sgt to the credential cache file.");}}
>  {{        boolean isFreshNew = !ccacheFile.exists();}}
>  {{        if (isFreshNew) {}}
>  {{            createCacheFile(ccacheFile);}}
>  {{        if (ccacheFile.exists() && ccacheFile.canWrite()) {}}
>  {{            }}
>  {{            try {}}
>  {{                CredentialCache cCache;}}
>  {{                if(!isFreshNew){}}
>  {{                    cCache = new CredentialCache(sgtTicket); 
> {color:#ff0000}//This constructor sets also the cCache principal from 
> sgtTicket principal{color}}}
>  {{                    cCache.load(ccacheFile);}}
>  {{                    cCache.addCredential(new Credential(sgtTicket, 
> sgtTicket.getClientPrincipal()));}}
>  {{                    
> cCache.setPrimaryPrincipal(sgtTicket.getClientPrincipal());}}
>  {{                } else {}}
>  {{                    cCache = new CredentialCache(sgtTicket);}}
>  {{                cCache.store(ccacheFile);}}
>  {{            } catch (IOException e) {}}
>  {{                throw new KrbException("Failed to store sgt", e);}}
>  {{        } else {}}
>  {{            throw new IllegalArgumentException("Invalid ccache file, "}}
>  {{                    + "not exist or writable: " + 
> ccacheFile.getAbsolutePath());}}
> Please note that *YOUR CredentialCache contructor assumes the clientPrincipal 
> is populated* ;)
> Hope useful,
> regards
> Fabiano
> P.s: Attached two paches for the project, bug2.diff for bug 2 is..*I think.. 
> production ready*.
> bug1.diff for bug 1 is untested.. I ported the modification of my overriding 
> method to the *AbstractInternalKrbClient.requestSgt()* method. But I have not 
> performed run/test of modules, your project is big and needs time to manage 
> well. *BUT* is this the right solution? or SHOULD the *clientPrincipal* be 
> already populated from the doRequestSgt method? I suggest to look for a 
> solution at a deeper level starting from *DefaultInternalKrbClient*. Consider 
> bug1.diff a fair workaround.
> P.s.2:Added a pull request on GitHub
> Regards



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to