so, if I do a pr_Initialize after the initial cell connect with no configuration directory, it resets enough of those that it connects to the correct ptserver (although it spews an error about no config file being found). Ugh.
the two solutions I see to fixing pr_initialize at the moment are:
1. make something not static so stuff isn't cached (I'm guessing it is being cached to try and speed up certain operations?)
2. add a litle bit of code to the top of the function to allow calling it that will reset a cached structure if, say, an empty confdir is passed (so that an error message isn't spewed to console everytime someone obtains tokens).
Thoughts?
On 2003.12.10 11:57 David Botsch wrote:
So, I've added the following two lines to before the call to ka_UserAuthenticateGeneral:
pr_End(); pr_Initialize(0, AFSDIR_CLIENT_ETC_DIRPATH, cell);
it still talks to ptservers for cell1 when attempting to resolve uid for cell2.
On 2003.12.10 11:44 Derrick J Brashear wrote:On Wed, 10 Dec 2003, David Botsch wrote:
> Maybe it's a bug, or maybe it is me doing something stupid. > > When authenticating users in more than one cell (ie enter uname/pw for > cell1, then later on, enter uname/pw for cell2 without exiting the auth > program), it seems that something someplace is caching and talking to > the wrong ptserver.
Yup. You need to call pr_End, and pr_Init again. (Actually, I wonder
if
pr_End works now; Some time ago it did not, but that was like AFS 3.4)
> When the uid lookup happens for cell2, it still talks to the ptservers > for realm1. _______________________________________________ OpenAFS-devel mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-devel
-- ******************************** David William Botsch Consultant/Advisor II CCMR Computing Facility [EMAIL PROTECTED] ******************************** _______________________________________________ OpenAFS-devel mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-devel
-- ******************************** David William Botsch Consultant/Advisor II CCMR Computing Facility [EMAIL PROTECTED] ******************************** _______________________________________________ OpenAFS-devel mailing list [EMAIL PROTECTED] https://lists.openafs.org/mailman/listinfo/openafs-devel
