Re: how to deal with hosts that are down

Brandon S. Allbery KF8NH (allbery@kf8nh.apk.net)
Tue, 22 Sep 1998 20:48:25 -0300


In message <5ld88n26rr.fsf@tequila.systemsz.cs.yale.edu>, Stefan Monnier
writes
:
+-----
| resolv for instance cycles through its listed servers, so if your first
| server is down, it will switch to the next after a time out. *But* this
| time out will be suffered for all applications over and over.
+--->8

Sun beat you to it; it's called nscd. I understand there's a version for
Linux; I just hope it works better than Sun's version (which in 2.5.1 at
least is seriously braindamaged[1]).

The basic idea is that nscd is a minimal caching nameserver for all
directory services (on Suns, this means it does DNS, YP/NIS, and NIS+).
Because it does the lookups instead of each program doing so itself, it can
remember state about e.g. nameservers being down.

[1] Maybe nscd is fine but those silly doors[2] are broken? We had name
lookups hanging all over the place; I finally truss'ed a program that was
waiting on a name lookup and it instantly unlocked and went on normally...
went on from door_return().

[2] Someone explain to me why Sun needed to invent yet another kind of IPC,
as if sockets, message queues, and two kinds of shared memory aren't enough?

-- 
brandon s. allbery	[os/2][linux][solaris][japh]	 allbery@kf8nh.apk.net
system administrator	     [WAY too many hats]	   allbery@ece.cmu.edu
electrical and computer engineering					 KF8NH
carnegie mellon university

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/