DB2 DPF
Linux
V 9.1


Hi,
I am trying to do a "df -h" on all the nodes in the cluster. I am running this script from the admin node.

cat ~/sqllib/db2nodes.cfg | awk '{print $2}'| uniq | while read nodename
do ssh $nodename "df -h" | grep -i % | cut -c 22- | awk '{print $5}' | grep -iE 'db2backups|db2containers|db2logs'
done

db2nodes.cfg has 3 nodes: adm, node1 and node2.

My above command is only displaying the df -h output of adm node and its not displaying the df -h outputs of node1 and node2.

If anyone could help in throwing a suggestion in this regard would be greatly appreciated.

Hi,

the script seem to be ok.

Did you check
cat ~/sqllib/db2nodes.cfg | awk '{print $2}'| uniq | while read nodename
do ssh $nodename "df -h"
done

Is the ssh generating the output for the other nodes? Are there keys for login and are host in the known_hosts?

Cheers
nvk

Why don't you use rah, the script that is supplied with DB2 exactly for this purpose?

--> nvk@vhv : There is no issue with ssh. When I try to execute the script mentioned by you, It generates df -h out put of only one node

But when I do,

cat ~/sqllib/db2nodes.cfg | awk '{print $2}'| uniq | while read nodename
do echo $nodename
done

The above one generates all the three nodenames. I am not sure why ssh on all the three nodes is not working for while loop

I modified while loop to for loop and then ssh started working on three nodes

What I did was

cat ~/sqllib/db2nodes.cfg | awk '{print $2}'| uniq > nodefile.out
for nodename in `cat nodefile.out`
do ssh $nodename "df -h" | grep -i % | cut -c 22- | awk '{print $5}' | grep -iE 'db2backups|db2containers|db2logs'
done

The above script works perfectly

--> n_i

Can we use rah to a specific node ? The reason why I am using ssh instead of rah is because, after doing ssh on each node, it generates an output specific to that node

This is what I am trying to do. If the particular file is not readable on that node, the output should say that the file is not readable on that specific node.

If you could suggest me a way how this could be accomplished through rsh, it would be really great.


cat ~/sqllib/db2nodes.cfg | awk '{print $2}'| uniq > nodefile.out
for nodename in `cat nodefile.out`
do ssh $nodename "df -h" | grep -i % | cut -c 22- | awk '{print $5}' | grep -iE 'db2backups|db2containers|db2logs' > filename_output
for filename in `cat filename_output`
do ssh "ls -ld $filename" | awk '{print $1}' | cut -c2 | while read var
if [[ $var -eq r ]]
then echo "$filename on $nodename is readable"
else
echo "$filename on $nodename is not readable"
fi
done
done