For what it's worth, I updated ACCUMULO-4150 to reflect the fact that I finally got the tests to pass with a newer version of Hadoop, so I'm satisfied now. :) That was really bugging me. My +1 stands.
On Thu, Feb 25, 2016 at 12:34 PM Christopher <[email protected]> wrote: > Oh, yes, apologies if I gave that impression. I'm sure we'll figure this > out, and if it is a problem in Accumulo's Kerberos feature (and not > something stupid on my end), I'm sure we're committed to fixing it quickly > and having it in the next bugfix release. > > On Thu, Feb 25, 2016 at 12:31 PM Josh Elser <[email protected]> wrote: > >> Thanks. I've been doing a bit with the Kerberos stuff (in real >> environments) and what is in 1.7.1 seems pretty solid to me. I just >> wanted to make sure people didn't avoid it, thinking that it was not >> stable. >> >> Christopher wrote: >> > Yeah, that's fine. I'll just reserve judgement for now, and defer to >> > others, since I don't have time to set up a separate Kerberos >> environment. >> > As far as I can tell, the feature works fine, up to a point. I'm just >> not >> > sure what to make of this particular test. As you said, though, it's >> quite >> > possibly just MiniKDC instabilities. >> > >> > On Thu, Feb 25, 2016 at 12:10 PM Josh Elser<[email protected]> >> wrote: >> > >> >> Welcome to why people say "Kerberos is hard". >> >> >> >> I think I said in chat, but increasing the timeout factor is not going >> >> to make that test pass if it can't pass the first time. The MiniKDC the >> >> tests use are not representative of a real KDC. I'd ask that you deploy >> >> Accumulo with Kerberos before passing judgement on the feature as a >> whole. >> >> >> >> I still have your IT logs -- I didn't get a chance to look at them >> >> yesterday. I'll try to do so today. >> >> >> >> Christopher wrote: >> >>> I had a lot of difficulty getting the Kerberos ITs to pass without >> timing >> >>> out. I was never able to get the KerberosRenewalIT to pass, even after >> >>> re-running several times (still trying), and even with a timeout >> factor >> >> of >> >>> 20. I do not have a strong confidence in the quality of the Kerberos >> >>> features as is, so it's not a critical feature for me, so I'll defer >> to >> >>> others tests for that. >> > >> >
