DNS "chicken-and-egg" Problem

[JINMEI Tatuya / 神明達哉]

At Mon, 27 Oct 2008 13:23:20 -0700 (PDT),
Fr34k <freaknetboy at yahoo.com> wrote:

> Below, there is a statement:
> "If it's occasional, that may be a known bug in 9.5 about
> cache entry management, and will be fixed in the next beta"
> 
> I am testing 9.5.1b2.
> 
> Is 9.5.1b2 affected by this "known bug"?

Yes, but note that there's a workaround as I just said in a separate
reply: raise max-cache-size.  The next beta will also "fix" the
inappropriate cache size problem.

> If so, do we have any details on this bug as well as any ETA on the "will be fixed in the next beta"?

We're now doing internal test for the next beta (9.5.1b3).  I don't
know the exact schedule, but I guess it will be available in a week or
so.

Regarding "details"..., I'm not sure how much you want to know, but in
short, this bug could purge cached entries too aggressively under an
'over memory' condition per max-cache-size.  In the worst case,
inserting a glue address record could purge the corresponding NS
record of the same (like the case of 'igpp.ucla.edu'), which
effectively kills the ongoing resolution process.

> Do you recommend downgrading to 9.5.1b1?

No, 9.5.1b1 has the same problem.

---
JINMEI, Tatuya
Internet Systems Consortium, Inc.