Mailinglist Archive: opensuse-security (601 mails)

< Previous Next >
NFS vs. cronjobs
  • From: Christian Mudra <mudra@xxxxxxxxxxxxxxxxxxxx>
  • Date: Fri, 4 Aug 2000 14:44:59 +0200
  • Message-id: <20000804144459.D30957@xxxxxxxxxxxxxxxxxxxxxxxxx>
Hi,

I'm running kernel 2.2.16 on SuSE 6.4. As NFS-server I'm using
the kernel-based nfsd.

Every night at 0:00 o'clock, the cron-jobs of SuSE are running both on
the nfs-clients and on the server.
Every night, beginning at 0:15 a.m. I get every 15 min a mail from each
client with the content (same like on console):
YPBINDPROC_DOMAIN: Domain not bound

I checked all the scripts that are running (/usr/lib/cron/run-crons,
/usr/lib/secchk/*, /etc/cron.daily/aaa_base*),but there I see no influence
to NFS.

Logging in to a client takes a lot (5++ min) of time. After a short time
or work (things that cause network traffic), everything is okay:
64 bytes from 131.246.*.*: icmp_seq=120 ttl=255 time=1000.802 ms
64 bytes from 131.246.*.*: icmp_seq=123 ttl=255 time=1307.996 ms
64 bytes from 131.246.*.*: icmp_seq=124 ttl=255 time=308.237 ms
64 bytes from 131.246.*.*: icmp_seq=125 ttl=255 time=2038.028 ms
64 bytes from 131.246.*.*: icmp_seq=126 ttl=255 time=1038.427 ms
64 bytes from 131.246.*.*: icmp_seq=127 ttl=255 time=38.693 ms
64 bytes from 131.246.*.*: icmp_seq=128 ttl=255 time=9317.865 ms
64 bytes from 131.246.*.*: icmp_seq=129 ttl=255 time=8318.299 ms
64 bytes from 131.246.*.*: icmp_seq=130 ttl=255 time=7318.521 ms
64 bytes from 131.246.*.*: icmp_seq=131 ttl=255 time=6318.699 ms
64 bytes from 131.246.*.*: icmp_seq=132 ttl=255 time=5318.947 ms
64 bytes from 131.246.*.*: icmp_seq=133 ttl=255 time=4319.162 ms
64 bytes from 131.246.*.*: icmp_seq=134 ttl=255 time=3319.374 ms
64 bytes from 131.246.*.*: icmp_seq=135 ttl=255 time=2319.922 ms
64 bytes from 131.246.*.*: icmp_seq=136 ttl=255 time=1320.099 ms
64 bytes from 131.246.*.*: icmp_seq=137 ttl=255 time=320.298 ms
64 bytes from 131.246.*.*: icmp_seq=138 ttl=255 time=0.270 ms
64 bytes from 131.246.*.*: icmp_seq=139 ttl=255 time=0.262 ms
64 bytes from 131.246.*.*: icmp_seq=140 ttl=255 time=0.254 ms

This ping was started directly after such a login - so you see, after
about 2 1/2 min of just being logged in the network is back to normal.
Logging in to such a slowed down machine is only possible as root,
not as user (user authentification uses NIS). But as security policy
we wanted to allow root login only via su, not remote, not on console.
As long as this problem exists, at least root login on console must
be allowed ...

Additionally sometimes there are messages like that in the syslog:
Aug 4 11:44:47 mycomp kernel: nfs: task 1674 can't get a request slot
Aug 4 11:44:49 mycomp kernel: nfs: task 1681 can't get a request slot
Aug 4 11:44:59 mycomp kernel: nfs: server OK

So, my question is: How can I avoid these nfs-problems? Has somebody
also experienced problems like this? And why does that begin on every
client just after the cronjobs finished? Ideas?

Greetings,

Christian.

----------------------------------------------------------------------------
Christian Mudra Just remember what your mother always told you:
mudra@xxxxxxxxxxxxxxxxxxxx "When you open windows you let in bugs".
AG Wiehagen, Universit├Ąt Kaiserslautern

< Previous Next >
Follow Ups