tty0 Diagnostics
9.3 Communication problems between PC and CCU
If the acquisition (zg, gs, wobb) and/or the configuration (cf) fail, you might have a communication problem between the LINUX PC and the spectrometer CCU. Typically you will get error messages from XWIN-NMR like:
Connect to spect: unable to connect to remote host
Connection to spect. aqport0, failed
Acquisition hardware not present or defect
Check the communication between the SGI and the CCU in the following way:
a) Open a shell
b) Type telnet spect
Depending on whether ‘telnet spect’ works (you get a login prompt) or not, continue with paragraph 9.3.1 or 9.3.2.
9.3.1 Communication problems but ‘telnet spect’ works
If ‘telnet spect’ works, the CCU has booted. This means that the spectrometer network software is installed and running. The communication problems obviouslyhave other causes. Two of the most common causes are mentioned below.
In each of the following steps you must be logged in on the SGI as root and type the indicated commands in a shell.
1. Check if the process startd is running on the CCU:
a) Type telnet spect and log in on the CCU as root
b) Type ps -efw| grep startd
Does this command show you 2 processes /etc/startd?
YES? Then startd is running, type exit, go to step 2
NO? Then you found the problem, go to step 1c.
c) Type /etc/startd
Trying to start startd in this way might show you the reason why startd was not started automatically. You might have a hardware problem and starting the startd manually indicates which board has a failure. Run the corresponding hardware test (see chapter 9.7.1).
2. Check if the spectrometer has an alias name which is not specified in /etc/hosts:
a) Type cd /<XWIN_NMR_HOME>/conf/instr
<XWIN_NMR_HOME> is the directory where XWIN-NMR is installed (default is /opt/xwinnmr)
b) Type more curinst
What is the output of this command?
• spect? Then the spectrometer does not have an alias name
• different from spect, e.g. drx ? Then do the following:
type grep spect /etc/hosts
What is the output of this command?
149.236.99.99 spect drx? Then the alias name is specified
149.236.99.99 spect? Then you found the problem, go to c)
c) Edit the file /etc/hosts:
append the alias name, e.g. drx after spect
If these two steps did not solve your problem, switch on the XWIN-NMR history (see chapter 9.4.3) to track all error messages and contact your local Bruker office.
9.3.2 Communication problems and ‘telnet spect’ does not work
If ‘telnet spect’ does not work, the CCU has probably not booted up completely. This can have several causes. In each of the following steps you must be logged in on the SGI as root and type the indicated commands in a shell. After each step, type telnet spect and see if you get a login prompt.
9.3.2.1 Software related problems
1. Become superuser; type su -2.
2. Reboot the SGI and CCU using reviveccu (The command reviveccu executes a shell script which will first remove the two files: /usr/diskless/clients/spect/root/etc/mtaband …/utmp and then reboot spectrometer-CCU and LINUX-PC) :
type reviveccu
.follow the instructions:
a) You will be asked to switch off the acquisition rack
b) reviveccu will automatically reboot the SGI
c) Switch on the acquisition rack after the SGI has rebooted
3. Check the nameserver configuration:
a) Type grep hosts /etc/nsswitch.conf
What is the output of this command?
Nothing? No problem, go to step 4.
• Cannot open nsswitch.conf ?No problem, no nameserver; go to step 4.
• hosts files ...? No problem, this entry is correct; go to step 4
• hosts dns…. or hosts nisplus….? You found the problem: go to step b.
b) Edit the file /etc/nsswitch.conf and change the line hosts so that the first entry after hosts is files.
4. Check if NFS is running:
a) Type ps -efw| grep nfsd
Does the output of this command show you 4 processes /usr/etc/nfsd ?
YES? Then NFS is running, continue with step 6
NO? Then you found the problem, continue with step 4b
b) Type /sbin/chkconfig nfs on
c) Type reviveccu
d) Run step 4a again and if NFS is still not running continue with step 5
5. Check if NFS is installed:
a) Type rpm -q nfs-utils
Does the output of this command show you the package nfs-utils?
YES? Then NFS is installed, go to step 6
NO? Then you found the problem, install the package nfs-utils. For a installation description please refer to chapter 2.4.1.1
6. Check if the process bfsd is running:
a) Type ps -efw| grep bfsd
Does the output of this command show you the process
/usr/diskless/bfsd.linux
YES? Then bfsd is running, continue with step 7.
NO? Then you found the problem, continue with step 6b
b) Check configuration of bfsd correct it if it is wrong
• type in /sbin/chkconfig –list bfsd and check if the output show ’on’ in runlevel 5, e. g.(If a process like this is not started, start it on a shell by typing /sbin/chkconfig <process>on) :
bfsd 0:off 1:off 2:on 3:on 4:on 5:on 6:off
c) Check the output of the command hostname.
Is the output of this command:
a plain hostname? No problem, continue with step 7.
a hostname.domainname? You found the problem. Change the hostname as described in chapter 3.2.3. Then continue with step 6d.
d) Type reviveccu
7. Check if the /usr/diskless partitions are exported:
a) Type /usr/sbin/exportfs -v
Does the output of this command show you both partitions with correct options:
/usr/diskless/dl_usr
spect (ro,async,wdelay,root_squash)
and /usr/diskless/clients/spect
spect (rw,async,wdelay,no_root_squash)
YES? Then the partitions are exported, continue with step 8
NO? Then you found the problem, continue with step 7b
b) Check the file /etc/exports and correct it if it is wrong (see chapter 9.3.3.1)
c) Type reviveccu
8. Check if the spectrometer network is correctly configured:
Please refer to chapter 3.2.1
9. Reconfigure the spectrometer network:
• You can use the automatic procedure which will be started by typing:
/<XWIN_NMR_HOME>/prog/bin/install.net/install.net
where <XWIN_NMR_HOME> is the directory where XWIN-NMR is installed (default /u). This command reconfigures the entire spectrometer network.
• You can check it manually, see chapter 3.2
10. Reconfigure the spectrometer network by re-installing the DISKLESS package:
Insert the XWIN-NMR CD into the CD drive, select the Diskless package and deselect all other packages.For more details see installation description in chapter 5.2
9.3.2.2 Hardware related problems
1. Check if the ethernet cable between SGI and CCU is properly connected.
.Remove and re-connect the ethernet cable and the transceiver or minihub.
2. Check if the ethernet interface is broken.
a) Physically connect the CCU to the first ethernet interface (ec0) (assuming that the CCU is currently connected to the second ethernet interface)
b) Become superuser; type su –
c) Change the configurations between eth0 and eth1 as described in chapter 3.1.2 and 3.2.1.
Note: Write down all configuration parameters like IP addresses, nameserver etc.
3. Replace the ethernet cable, transceiver or minihub between the PC and CCU to find out if either of these components is broken.
4. If a new CCU was installed:
a) Type arp -d spect
b) Type reviveccu
9.3.3 Files which are needed for the spectrometer network
9.3.3.1 File entries made for the spectrometer network
• /etc/hosts
149.236.99.1 ASP_ST2
149.236.99.99 spect
• /etc/bootparams
spect root=ASP_ST2:/usr/diskless/clients/spect/root
swap =ASP_ST2:/usr/diskless/clients/spect/swapfile
dump= ASP_ST2:/usr/diskless/clients/spect/dump
Note that this is one line with a blank between the three entries. In older versions of the NMR Suite, these were three separate lines with a backslash (\) at the end of each line.
• /etc/exports
/usr/diskless/dl_usr spect(ro)
/usr/diskless/clients/spect spect(rw,no_root_squash)
• /etc/sysconfig/network
NETWORKING=yes
HOSTNAME=hosttest
GATEWAYDEV=
GATEWAY=
• /etc/sysconfig/network-scripts/ifcfg-eth1
BROADCAST=149.236.99.255
NETWORK=149.236.99.0
NETMASK=255.255.255.0
IPADDR=149.236.99.1
DEVICE=eth1
ONBOOT=yes
BOOTPROTO=none
USERCTL=no
• /usr/diskless/clients/spect/root/etc/fstab
ASP_ST2:/usr/diskless/clients/spect/root / nfs rw,bg,hard,retry=3,timeo=20 0 0
ASP_ST2:/usr/diskless/dl_usr /usr nfs ro,bg,hard,retry=3,timeo=20 0 0
ASP_ST2:/usr/diskless/clients/spect/var /var nfs rw,bg,hard,retry=3,timeo=20 0 0
• /etc/xinetd.d/bootparamd
service bootparamd
{
type= RPC
socket_type= dgram
user= root
server= /usr/sbin/rpc.bootparamd
wait= yes
protocol= udp
rpc_version= 1
disable= no
instances= 1
log_type= FILE /usr/diskless/bootparamd.log
}
9.3.3.2 Files created for the spectrometer network
• /usr/diskless/clients/spect/swapfile
This file can also be created manually with the command:
dd if=/dev/zero of=/usr/diskless/clients/spect/swap-file bs=8000000 count=1
• /etc/rc2.d/S95bfsd
This file is a symbolic link to the file /etc/init.d/bfsd. From this file the bfsd daemon /usr/diskless/bfsd is started.
• /usr/diskless/clients/spect/root/dev
This directory contains a large number of special files such as tty00, tty01, tty02, etc. These files are created during the installation of the DISKLESS package from the XWIN-NMR CD. They can also be created manually:
a) Become superuser: type su –
b) cd /usr/diskless/clients/spect/root/dev
c)./MKDEV -m RS3330
View the output of the CCU console
In order to view the output of the CCU you must connect serial port 2 of the PC with tty00 of the CCU, you need a cable with a round rs232 plug on the CCU side (order number HZ04161/A). The program cu can be used to make the output of the CCU visible in a shell on the PC.
Check if cu is installed:
rpm -q uucp
if it is installed and active the answer should be:
uucp-<versions-no>
If it is not installed yet, please refer to the installation description in chapter 2.4.1.2
In case of any problems with this package check the permissions of the program.
Type ls -l /dev/ttyS1
The permissions, owner, and group of this file must be: crw-rw—-, root and uucp respectively. If they are different, change them as follows:
chmod 660 /dev/ttyS1
chown root.uucp /dev/ttyS1
Type /sbin/reboot
9.3.4.1 View/change the prom parameters
1. Open a shell
2. Type cu -lttyS1
.You will get the prompt connected and you can see the CCU output messages:
a) Push the reset button of the CCU
b) Press Control-C to interrupt the boot process
The CCU will be in Monitor Mode now, the prompt will change to >>, type:
>> printenv
You will get a list of environment variables, the most important are:
netaddr=149.236.99.99
netmask=0xffffff00
bootfile=bfs()/usr/diskless/clients/
spect/root/unix_r4600_std
bootmode=c
These values must be shown as specified above ( If the netaddr is different, the spectrometer network files must be adjusted. However, when you install the DISKLESS package the entries are set back to 149.236.99.99 and your CCU will not boot anymore. Therefore we strongly recommend not to change the netaddr) . You can set the environment variables as follows:
>> setenv netaddr 149.236.99.99
>> setenv netmask 0xffffff00
To boot the CCU again, type:
>> u
c) To exit the cu program type:
>> ~.Enter (press the keys: tilde, dot, Enter)
9.3.4.2 Viewing the boot process
When all environment parameters are correct, you can view the boot process and see where it hangs. This might give you an indication as to what the problem is.
1. Open a shell
2. Type cu -lttyS1
.you will get the prompt connected and after typing Enter you can see the CCU output
3. Push the reset button of the CCU
The CCU will try to boot now. The output of the boot process is listed below. The marks ‘—?—’ indicate positions where the boot process possibly hangs or goes into an endless loop:
Autoboot: Waiting to load
bfs()/usr/diskless/clients/spect/root/unix.r4600_std
(CTRL-C to abort)
loading
—1–
Obtaining /usr/diskless/clients/spect/root/unix.r4600_std from
server c85
851056+115728+824448 entry: 0x80021000
CPU: MIPS R4700 Processor Chip Revision: 1.0
FPU: MIPS R4700 Floating Point Unit [CP1] Revision: 1.0
RISC/os Release 4_52 mips Version UMIPS
Total real memory = 16777216
Available memory = 14356480
—2––
Root on nfs file : Swap on nfs file :
—3––
hostname: spect
domainname: dummy
—4––
Root fstype nfs
Available memory = 12705792
—5––
Root on nfs file : Swap on nfs file
—6––
The system is coming up. Please wait.
ASP_ST2:/usr/diskless/clients/spect/var mounted on /var
ASP_ST2:/usr/diskless/dl_usr mounted on /usr
Internet daemons: portmap inetd.
NFS daemons: nfsd biod lockd statd.
The system is ready.
Booting might hang, stop (and print an error message) or go into an endless loop at one of the positions 1 through 5. This might be caused by one of the problems mentioned below. Check the indicated files and correct them if necessary or perform the indicated steps.
a) If booting stops at position —1— you might have one of the following prob-lems:
• the bfsd is not running; go to 9.3.2.1, step 6
• the DISKLESS packages is not installed; see chapter 9.3.2.1, step 10
• the network is not set up correctly; check the configuration (see chapter 3.2.1)
If the problems above do not exist, please enter the following command in a shell:
/sbin/arp -a
The output should contain a line like:
spect (149.236.99.99) at a.b.c.d.e.f [ether] on eth1
where a.b.c.d.e.f is a hexadecimal hardware ethernet address of the CCU.
b) If booting stops at position —2— please try the following:
• push the CCU reset button
• type Ctrl-C to interrupt the boot process and enter:
>> setenv bootmode m
>> setenv bootmode c
• push the CCU reset button
c) If booting stops at position —3— one of the following problems exist:
• the CCU (hostname spect) is searched for on a nameserver first:
The search order is determined by hosts:
/etc/nsswitch.conf: hosts files…
Make sure the first location to be searched is the local /etc/hosts file as specified by the entry files as specified above.
• the network is not set up correctly; check /etc/hosts, /etc/boot-params
• the swapfile does not exist; check the file
/usr/diskless/clients/spect/swapfile
if the file does not exist please refer to chapter 9.3.3.2
• start of bootparam fails; check the entry bootparamd:
/sbin/chkconfig –list bootparamd
if the entry for 5 is off activate it:
/sbin/chkconfig –level 5 bootparamd on
• rename the file ’/etc/hosts.deny’ to ’/etc/hosts.deny.old’
d) If booting stops at position —4— one of the following problems exist:
• NFS is not running; see chapter 9.3.2.1, step 4
• diskless partitions are not exported; see chapter 9.3.2.1, step 7
• the network is not set up correctly; check the file /etc/hosts
e) If booting stops at position —5— or —6— with the following error message please do like describe:
mount root ASP_ST2:/usr/diskless/clients/spect/root
failed, rpc status 15
PANIC: vfs_mountroot: cannot mount root’you had to check the file ( see chapter 9.3.3.1)
/etc/xinetd.d/bootparamd
f) If booting is very slow or hangs at an arbitrary point you might have a hardware problem (see chapter 9.3.2.2).
4. To exit the cu program type:
>> ~.Enter (press the keys: tilde, dot, Enter)
9.3.4.3 Running CCU diagnostic tests
1. Open a shell
2. Type cu -lttyS1
.you will get the prompt connected and you can see the CCU output:
a) Push the reset button of the CCU
b) Press Control-C to interrupt the boot process
The CCU will be in Monitor Mode now, the prompt will change to >>
>> setenv bootmode m
c) Push the CCU reset button
The CCU diagnostic tests will start automatically. If one or more diagnostic tests fail, the CCU could be broken. If they are all passed successfully, your CCU seems to be okay. The bootmode is automatically set back to its normal value c.
d) To exit the cu program type:
>> ~.Enter (press the keys: tilde, dot, Enter)