Anything with NISPLUS takes forever
Christian Leber
christian at leber.de
Wed Oct 29 23:01:52 EST 2003
Hello,
I have here some Debian GNU/Linux boxes that are using 2 (master,
replica) Solaris 8 Servers.
Mostly everything[1] works, but exactly now there is something really
obscure going on:
Box A:
sunburn:~# time id christian
uid=1097(christian) gid=30(students)
groups=30(students),10(staff),22(develop),29(atollusr)
real 0m7.404s
Box B:
nala:~# time id christian
uid=1097(christian) gid=30(students)
groups=30(students),10(staff),22(develop),29(atollusr)
real 0m0.824s
7 seconds is really a little bit too long...
Both have exactly the same relevant packages (glibc, libxcrypt,
libpam-unix2, niscplus-client, kernel) and exactly the same /etc/nsswict.conf,
/etc/pam.d etc. and are at the same switch.
It is Debian woody, therefore glibc 2.2.5.
The nisplus servers are pinagble from both hosts.
I did a strace on both boxes, on A it is 680 kb and on B it is 32 kb in
size.
First difference:
Box A:
bind(3, {sa_family=AF_INET, sin_port=htons(988), sin_addr=inet_addr("0.0.0.0")}, 16) = 0
ioctl(3, FIONBIO, [1]) = 0
Box B:
bind(3, {sin_family=AF_INET, sin_port=htons(778), sin_addr=inet_addr("0.0.0.0")}}, 16) = 0
ioctl(3, 0x5421, [1]) = 0
Everything takes forever, has somebody got an idea?
thanks you in advance, slowly this gets really mad here
Regards
Christian Leber
[1] well, logging in with public key doesn't work without logging in one
time before with password...
P.S. how hard is it to change the solaris boxes to ldap?
--
"Omnis enim res, quae dando non deficit, dum habetur et non datur,
nondum habetur, quomodo habenda est." (Aurelius Augustinus)
Translation: <http://gnuhh.org/work/fsf-europe/augustinus.html>
More information about the linux-nisplus
mailing list