As stated in my email, my client and server are the same machine.
Here are the permissions on .amandahosts:
-rw------- 1 backup disk 174 May 22 13:14
/root/amanda-data/.amandahosts
Adding root entries, like this to .amandahosts:
10.1.0.2 backup
10.1.0.3 backup
server02.ct backup
server01.ct backup
broadcast backup
10.1.0.2 root
10.1.0.3 root
server02.ct root
server01.ct root
broadcast root
did not change any behavior. The debug/log output is identical.
amcheck still returns:
WARNING: server02.ct: selfcheck request timed out. Host down?
Client check: 1 host checked in 30.036 seconds, 1 problem found
Thank you,
Cory Visi
"Martinez, Michael -
CSREES/ISTM" To: 'Cory Visi'
<[EMAIL PROTECTED]>,
<[EMAIL PROTECTED] [EMAIL PROTECTED]
eusda.gov> cc:
Subject: RE: Me again, still
getting
05/22/2002 12:29 PM WARNING: host: selfcheck request
time d out.
Host down?
You're .amandahosts on your tape server needs to include a line for "root"
for all your tape clients (including itself) ; and .amandahosts on your
clients needs to be chmod 400
-----Original Message-----
From: Cory Visi [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, May 22, 2002 11:36 AM
To: [EMAIL PROTECTED]
Subject: Me again, still getting WARNING: host: selfcheck request timed
out. Host down?
Ok, I have looked into my issues further. Many replies on the list have
helped me, although I am still having problems and I do not understand what
Amanda is doing.
Firstly, this error:
amandad: dgram_send_addr: sendto(0.0.0.0.929) failed: Invalid argument
Is clearly not indicative of an error. Amanda uses broadcasts to find
clients/servers regardless of the configuration. Some people get this error
but have a perfectly working system.
This error, however:
Amanda Backup Client Hosts Check
--------------------------------
WARNING: server02.ct: selfcheck request timed out. Host down?
Client check: 1 host checked in 30.028 seconds, 1 problem found
IS indicative of a problem.
So, starting from scratch I will explain all the details of my setup:
I am using Amanda 2.4.2p2 on Linux 2.4.13 with glibc 2.2.2 compiled by
myself.
I am running amandad, amindexd, amidxtaped from xinetd. All these protocols
are listed in my /etc/services.
Currently, I am only trying to backup the data on the same machine as the
server, so the server and client are the _same machine_. I am using the
eth0 interface IPs to communicate, although, clearly the packets will go
through the loopback. The local firewall is configured to _allow all
packets_ through on the loopback and the eth0 interface (since it's local
only).
The following are relevant files concerning my setup. Some of my setup may
seem redundant and superfluous, however, in my efforts to make things work,
I have tried a number of things.
/etc/passwd entry for amanda user:
backup:x:417:6:Amanda Backup:/root/amanda-data:/bin/bash
/root/amanda-data/.amandahosts:
10.1.0.3 backup
server02.ct backup
broadcast backup
/etc/hosts:
127.0.0.1 localhost.localdomain localhost
10.1.0.3 server02.ct.internaldomain.com server02.ct
0.0.0.0 broadcast broadcast
some of amanda.conf:
dumpuser "backup"
inparallel 4
netusage 51200 Kbps
dumpcycle 3 weeks
runspercycle 15
tapecycle 20 tapes
runtapes 1
tpchanger "chg-manual"
tapedev "/dev/nst0"
changerfile "/root/amanda-data/changer-status"
changerdev "/dev/null"
tapetype DAT
infofile "/root/amanda-data/daily"
logdir "/var/log/amanda/daily"
indexdir "/root/amanda-data/daily"
tapelist "/root/amanda-data/daily/tapelist"
define tapetype DAT {
comment "Archive Python 04687-XXX"
length 4000 mbytes
filemark 100 kbytes
speed 800 kbytes
}
disklist:
server02.ct sda1 comp-root
server02.ct sda2 comp-user
server02.ct sdb1 comp-user
xinetd entries:
service amanda
{
disable = no
bind = 10.1.0.3
log_on_success += USERID
log_on_failure += USERID
socket_type = dgram
protocol = udp
wait = yes
user = root
server = /usr/local/libexec/amandad
}
service amandaidx
{
disable = no
bind = 10.1.0.3
log_on_success += USERID
log_on_failure += USERID
socket_type = stream
protocol = tcp
wait = no
user = backup
server = /usr/local/libexec/amindexd
}
service amidxtape
{
disable = no
bind = 10.1.0.3
log_on_success += USERID
log_on_failure += USERID
socket_type = stream
protocol = tcp
wait = no
user = backup
server = /usr/local/libexec/amidxtaped
}
Following are my debug files for amcheck.
amcheck.debug:
amcheck: debug 1 pid 24151 ruid 417 euid 0 start time Wed May 22 10:29:30
2002
amcheck: dgram_bind: socket bound to 0.0.0.0.929
amcheck: pid 24151 finish time Wed May 22 10:30:00 2002
selfcheck.debug:
selfcheck: checking disk sdb1
selfcheck: device /dev/sdb1
selfcheck: OK
selfcheck: checking disk sda2
selfcheck: device /dev/sda2
selfcheck: OK
selfcheck: checking disk sda1
selfcheck: device /dev/sda1
selfcheck: OK
selfcheck: pid 8251 finish time Wed May 22 10:29:30 2002
amandad.debug:
got packet:
--------
Amanda 2.4 REQ HANDLE 000-48C30708 SEQ 1022077770
SECURITY USER backup
SERVICE selfcheck
OPTIONS ;
DUMP sdb1 0 OPTIONS |;bsd-auth;compress-fast;
DUMP sda2 0 OPTIONS |;bsd-auth;compress-fast;
DUMP sda1 0 OPTIONS |;bsd-auth;compress-fast;
--------
sending ack:
----
Amanda 2.4 ACK HANDLE 000-48C30708 SEQ 1022077770
----
amandad: dgram_send_addr: sendto(0.0.0.0.929) failed: Invalid argument
bsd security: remote host broadcast user backup local user backup
amandahosts security check passed
amandad: running service "/usr/local/libexec/selfcheck"
amandad: sending REP packet:
----
Amanda 2.4 REP HANDLE 000-48C30708 SEQ 1022077770
OPTIONS ;
OK /dev/sdb1
OK /dev/sda2
OK /dev/sda1
OK /sbin/dump executable
OK /bin/gzip executable
OK /etc/dumpdates read/writable
OK /dev/null read/writable
OK /tmp/amanda has more than 64 KB available.
OK /tmp/amanda has more than 64 KB available.
OK /etc has more than 64 KB available.
----
amandad: dgram_send_addr: sendto(0.0.0.0.929) failed: Invalid argument
You can see that amanda is communicating ONLY over the broadcast. Thanks,
and only thanks, to my manipulation of /etc/hosts and .amandahosts, this is
producing any output at ALL. If I remove the broadcast hosts and
.amandahosts entry, selfcheck doesn't even RUN.
In any case, this is how I interpret the above output (please correct me if
I'm wrong). The client is contacting the server over the broadcast address
0.0.0.0 as we have discovered amanda always does, however, that's it. It's
not contacting the server over a unicast address at all! amandad receives
the request, validates the user, and processes the request. Upon attempting
to send back the output, it fails, because you can not have one-to-one
communication on a broadcast address (i.e. Invalid argument).
Why does not the client try to use the unicast address I have specified in
the disklist? Even though the entries clearly resolve, I have tried using
the IP also with the same results. Why is amcheck only binding to 0.0.0.0
and not 10.1.0.3? What am I doing wrong?
Thank you,
Cory Visi
