On Tue, 2005-03-15 at 09:13 -0600, Zachary Miller wrote: > what you want to say is this: > > SEC_CLIENT_AUTHENTICATION = OPTIONAL > SEC_CLIENT_ENCRYPTION = OPTIONAL > SEC_CLIENT_INTEGRITY = OPTIONAL > > all outgoing connections look first at SEC_CLIENT_*, and fall back to > SEC_DEFAULT_* if the first is not found. this tells the client it doesn't > need to authenticate unless the server asks it to, which is what you want. > > so our collector wont ask you to authenticate your update, but your local > daemons will still require it. (they will use the SEC_DEFAULT_*) Gotcha. > acutally, all outgoing commands look at SEC_CLIENT_*. the other levels (READ, > WRITE, ADMINISTRATOR) are looked at only by the server side to enforce policy > on incoming commands. > > please let me know if that wasn't clear. No, that makes excellent sense. The main part that was throwing me off was that I thought that the SEC_WRITE_* fields must be being used as given that my logs didn't have any mention of ANONYMOUS access being attempted; but of course, as you don't have ANONYMOUS access permitted on your collector then the negotiation phase would have already eliminated that as an option. > side note: > > # Anonymous access is sufficient for read access. > > SEC_CLIENT_AUTHENTICATION_METHODS = KERBEROS, ANONYMOUS > > SEC_READ_AUTHENTICATION_METHODS = KERBEROS, ANONYMOUS > > you'll still be using KERBEROS for read commands. while both the client side > (which looks at SEC_CLIENT_*) and the server side (which looks at SEC_READ_*) > both have anonymous in the list, the list is ordered. you should say: > > SEC_READ_AUTHENTICATION_METHODS = ANONYMOUS, KERBEROS > > instead. it is the order of the methods that the server sees that is used. Yeah, that was actually deliberate. I wanted to allow anonymous access (so that my CondorStats[0] cronjob could get the current pool state without needing to keep TGT's around) whilst still using Kerberos as the default if it was available. That way end-user actions would be accountable by default, which I thought might be handy later. Thanks very much. Cheers, David -- David McBride <dwm@xxxxxxxxxxxx> Department of Computing, Imperial College, London
Attachment:
signature.asc
Description: This is a digitally signed message part