DNS - BIND 8.3.4-REL process dying with exit 0 status
Pasika, P M [NTK]
Pimp.M.Pasika at mail.sprint.com
Tue Aug 31 16:56:29 UTC 2004
Update:
Subject updated to reflect the actual problem more accurately.
I had fixed the error with stat'ing ndc. However named will still exit
out with a status 0. Here is the truss output:
fcntl(5, F_SETFL, 0x00000082) =3D 0
close(5) =3D 0
stat("/usr/local/etc/ndc.d/ndc", 0xFFBEFAC0) =3D 0
unlink("/usr/local/etc/ndc.d/ndc") =3D 0
llseek(0, 0, SEEK_CUR) =3D 0
_exit(0)
Here is the named.conf for that server. IPs modified for obvious
reasons. =20
options {
directory "/usr/local/etc/dns";
// forwarders { 248.4.96.5; 22.15.68.36; };
// forward only;
statistics-file "/var/log/named.ndcstats";
pid-file "/usr/local/etc/named.pid" ;
// recursion no;
// fake-iquery yes;
allow-transfer {
43.2.41.148; // Sacramento DNS
65.3.217.248; // Santa Clara DNS
87.7.41.248; // Independence DNS
129.6.41.148; // Omaha DNS
254.11.217.148; // Branchburg DNS
1.11.169.148; // Cheektowaga DNS
146.14.9.148; // Indianapolis DNS
27.15.137.68; // Lansing DNS
69.18.41.78; // Fort Worth DNS
145.18.201.78; // New Orleans DNS
234.26.41.28; // Charlotte DNS
210.26.169.18; // Norcross DNS
};
};
//controls {
// unix "/var/run/ndc" perm 0600 owner 0 group 0;
//};
logging {
channel my_syslog {
syslog daemon;
severity info;
};
channel stats_file {
file "/crash/named.stats" versions 10 size 200000k;
severity debug 3;
print-category yes;
print-time yes;
};
category default { my_syslog; };
category queries { stats_file; };
category notify { stats_file; };
category xfer-in { stats_file; };
category xfer-out { stats_file; };
};
acl "Internal" {10/8; 208.4.106/24; 127.0.0.1; };
zone "." in {
type hint;
file "db.cache";
};
zone "0.0.127.in-addr.arpa" in {
type master;
file "db.127.0.0";
};
I have commented out the "statistics-file "/var/log/named.ndcstats";"
line in the mean time to see what happens. Although, this breaks my
metric graphing. What are your thoughts?
-Paul Pasika
Sprint PCS
913-227-5097
paulpasika at nmcc.sprintspectrum.com
-----Original Message-----
From: bind-users-bounce at isc.org [mailto:bind-users-bounce at isc.org] On
Behalf Of Pasika, P M [NTK]
Sent: Saturday, August 28, 2004 3:52 AM
To: bind-users at isc.org
Subject: DNS - BIND 8.3.4-REL process dying without core on Solaris 7
Hello,b
I'm running BIND 8.3.4-REL on a Netra 1400 w/ Solaris 7 installed. I've
been noticing that named has been dying about every 2-5 minutes with no
pattern at all. I have a process monitoring script respawning it
seamlessly; however I am unable to identify the cause of this error. I
tried running a truss on the process to identify what is going on when
the process dies, below will be some pastes from 3 different
occurrences. It may have something to do with /usr/local/etc/ndc.d.ndc
being unable to stat. The permissions are:
# ls -l
total 0
srwxr-xr-x 1 root other 0 Aug 28 01:38 ndc
# file ndc
ndc: socket
The process is being run as user named, should I manually change the
owner of the file or leave it as is? (edit: I changed ownership to
named, to no avail.)
I know I need to update to 8.3.7, however the vulnerability of 8.3.4
should not be causing this to happen, I believe. I compiled with gcc
2.95 (I couldn't get it to compile with 3.x) and the box is serving
about 1 million queries/hour. I would like to identify the issue at
hand, upgrading to 9 seems prudent and unavoidable, however at this
time, this is my most pressing issue. Thanks for all of your help in
advance.
Attempt #1
getpid() =3D 9632 [1]
door_call(7, 0xFFBEE788) =3D 0
close(7) =3D 0
fstat(3, 0xFFBEF1B0) =3D 0
time() =3D 1093682128
getpid() =3D 9632 [1]
putmsg(3, 0xFFBEE868, 0xFFBEE85C, 0) =3D 0
open("/etc/.syslog_door", O_RDONLY) =3D 7
door_info(7, 0xFFBEE7A0) =3D 0
getpid() =3D 9632 [1]
door_call(7, 0xFFBEE788) =3D 0
close(7) =3D 0
fcntl(5, F_GETFL, 0x00000000) =3D 130
fstat64(5, 0xFFBEF7E0) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF8E0, 0xFFBEF8D8, -14572420) =3D 0
fcntl(5, F_GETFL, 0x00000000) =3D 130
fstat64(5, 0xFFBEF858) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF958, 0xFFBEF950, 267386896) =3D 0
fstat64(5, 0xFFBEF858) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF958, 0xFFBEF954, 267386896) =3D 0
setsockopt(5, 65535, 8192, 0xFFBEF958, 4, 267386896) =3D 0
fcntl(5, F_SETFL, 0x00000082) =3D 0
close(5) =3D 0
stat("/usr/local/etc/ndc.d/ndc", 0xFFBEFAC0) Err#13 EACCES
llseek(0, 0, SEEK_CUR) =3D 0
_exit(0)
Attempt #2
fstat64(5, 0xFFBEF858) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF958, 0xFFBEF954, 1609564176) =3D 0
setsockopt(5, 65535, 8192, 0xFFBEF958, 4, 1609564176) =3D 0
fcntl(5, F_SETFL, 0x00000082) =3D 0
close(5) =3D 0
stat("/usr/local/etc/ndc.d/ndc", 0xFFBEFAC0) Err#13 EACCES
llseek(0, 0, SEEK_CUR) =3D 0
_exit(0)
------------------------------------------------------------------------
------------------------------------------------------------------------
------------------------------------------------------------------------
---------------
Attempt #3
getpid() =3D 13900 [1]
door_call(7, 0xFFBEE788) =3D 0
close(7) =3D 0
fstat(3, 0xFFBEF1B0) =3D 0
time() =3D 1093682626
getpid() =3D 13900 [1]
putmsg(3, 0xFFBEE868, 0xFFBEE85C, 0) =3D 0
open("/etc/.syslog_door", O_RDONLY) =3D 7
door_info(7, 0xFFBEE7A0) =3D 0
getpid() =3D 13900 [1]
door_call(7, 0xFFBEE788) =3D 0
close(7) =3D 0
fcntl(5, F_GETFL, 0x00000000) =3D 130
fstat64(5, 0xFFBEF7E0) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF8E0, 0xFFBEF8D8, 0) =3D 0
fcntl(5, F_GETFL, 0x00000000) =3D 130
fstat64(5, 0xFFBEF858) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF958, 0xFFBEF950, 267386896) =3D 0
fstat64(5, 0xFFBEF858) =3D 0
getsockopt(5, 65535, 8192, 0xFFBEF958, 0xFFBEF954, 267386896) =3D 0
setsockopt(5, 65535, 8192, 0xFFBEF958, 4, 267386896) =3D 0
fcntl(5, F_SETFL, 0x00000082) =3D 0
close(5) =3D 0
stat("/usr/local/etc/ndc.d/ndc", 0xFFBEFAC0) Err#13 EACCES
llseek(0, 0, SEEK_CUR) =3D 0
_exit(0)
-Paul Pasika
Sprint PCS
913-227-5097
paulpasika at nmcc.sprintspectrum.com
More information about the bind-users
mailing list