I'm trying to write a script that builds a list of nodes then ssh into the first node of that list
and runs a checknodes.sh script which it's self is just a for i loop that calls checknode.sh
The first 2 lines seems to work ok, the list builds successfully, but then I get either get just the echo line of checknodes.sh to print out or an error saying cat: gpcnodes.txt: No such file or directory
MYSCRIPT.sh:
#gets the master node for the job
MASTERNODE=`qstat -t -u \* | grep $1 | awk '{print$8}' | cut -d'#' -f 2 | cut -d'.' -f 1 | sed -e 's/$/.com/' | head -n 1`
#builds list of nodes in job
ssh -qt $MASTERNODE "qstat -t -u \* | grep $1 | awk '{print$8}' | cut -d'#' -f 2 | cut -d'.' -f 1 | sed -e 's/$/.com/' > /users/issues/slow_job_starts/gpcnodes.txt"
ssh -qt $MASTERNODE cd /users/issues/slow_job_starts/
ssh -qt $MASTERNODE /users/issues/slow_job_starts/checknodes.sh
checknodes.sh
for i in `cat gpcnodes.txt `
do
echo "### $i ###"
ssh -qt $i /users/issues/slow_job_starts/checknode.sh
done
checknode.sh
str=`hostname`
cd /tmp
time perf record qhost >/dev/null 2>&1 | sed -e 's/^/${str}/'
perf report --pretty=raw | grep % | head -20 | grep -c kernel.kallsyms | sed -e "s/^/`hostname`:/"
When ssh -qt $MASTERNODE cd /users/issues/slow_job_starts/ is finished, the changed directory is lost.
With the backquotes replaced by $(..) (not an error here, but get used to it), the script would be something like
for i in $(cat /users/issues/slow_job_starts/gpcnodes.txt)
do
echo "### $i ###"
ssh -nqt $i /users/issues/slow_job_starts/checknode.sh
done
or better
while read -r i; do
echo "### $i ###"
ssh -nqt $i /users/issues/slow_job_starts/checknode.sh
done < /users/issues/slow_job_starts/gpcnodes.txt
Perhaps you would also like to change your last script (start with cd /users/issues/slow_job_starts)
You will find more problems, like sed -e 's/^/${str}/' (the ${str} inside single quotes won't be replaced by a host), but this should get you started.
EDIT:
I added option -n to the ssh call.
Redirects stdin from /dev/null (actually, prevents reading from stdin).
Without this option only one node is checked.
I am working on a bash script that uses pssh to run external commands, then join the output of the commands with the IP of each server. pssh has an option -o that writes a file for each server into a specified directory, but if the commands do not run, you just have an empty file. What I am having issues with is updating these empty files with something like "Server Unreachable" so that I know there was a connection issue reaching the server and to not cause problems with the rest of the script.
Here is what I have so far:
#!/bin/bash
file="/home/user/tools/test-host"
now=$(date +"%F")
folder="./cnxhwinfo-$now/"
empty="$(find ./cnxhwinfo-$now/ -maxdepth 1 -type f -name '*' -size 0 -printf '%f%2d')"
command="echo \$(uptime | awk -F'( |,|:)+' '{d=h=m=0; if (\$7==\"min\") m=\$6; else {if (\$7~/^day/) {d=\$6;h=\$8;m=\$9} else {h=\$6;m=\$7}}} {print d+0,\"days\",h+0,\"hours\",m+0,\"minutes\"}'), \$(hostname | awk '{print \$1}'), \$(sudo awk -F '=' 'FNR == 2 {print \$2}' /etc/connex-release/version.txt), \$(lscpu | awk -F: 'BEGIN{ORS=\", \";} NR==4 || NR==6 || NR==15 {print \$2}' | sed 's/ *//g') \$(free -k | awk '/Mem:/{print \$2}'), \$(df -Ph | awk '/var_lib/||/root/ {print \$2,\",\"\$5,\",\"}')"
pssh -h $file -l user -t 10 -i -o /home/user/tools/cnxhwinfo-$now -x -tt $command
echo "Server Unreachable" | tee "./cnxhwinfo-$now/$empty"
ls ./cnxhwinfo-$now >> ./cnx-data-$now
cat ./cnxhwinfo-$now/* >> ./cnx-list-$now
paste -d, ./cnx-data-$now ./cnx-list-$now >>./cnx-data-"$(date +"%F").csv"
I was trying to use find to locate the empty files and write "Server" unavailable using tee with this:
echo "Server Unreachable" | tee "./cnxhwinfo-$now/$empty"
if the folder specified doesn't already exist i get this error:
tee: ./cnxhwinfo-2019-09-03/: Is a directory
And if it does exist (ie, i run the script again), it instead creates a file named after the IP addresses returned by the find command, like this:
192.168.1.2 192.168.1.3 192.168.1.4 1
I've also tried:
echo "Server Unreachable" | tee <(./cnxhwinfo-$now/$empty)
The find command outputs the IP addresses on a single line with a space in between each one, so I thought that would be fine for tee to use, but I feel like I am either running into syntax issues, or am going about this the wrong way. I have another version of this same script that uses regular ssh and works great, just much slower than using pssh.
empty should be an array, assuming none of the file names will contain any whitespace in their names.
readarray -t empty < <(find ...)
echo "Server unreachable" | (cd ./cnxhwinfo-$now/; tee "${empty[#]}" > /dev/null)
Otherwise, you are building a single file name by concatenating the empty file names.
I am trying currently to achieve a bash script that will validate if SSH keys on a server are still linked to known hosts that are active on the local area network. You can find below the beginning of my bash script to achieve this:
#!/bin/bash
# LAN SSH KEYS DISCOVERY SCRIPT
# TRYING TO FIND THOSE SSH KEYS NOW
cat /etc/passwd | grep /bin/bash > bash_users
cat bash_users | cut -d ":" -f 6 > cutted.bash_users_home_dir
for bash_users in $(cat cutted.bash_users_home_dir)
do
ls -al $bash_users/.ssh/*id_* >> ssh-keys.txt
done
# DISCOVERING THE KNOWN_HOSTS NOW
for known_hosts in $(cat cutted.bash_users_home_dir)
do
cat $bash_users/.ssh/known_hosts | awk '{print $1}' | sort -u >>
hosts_known.txt
sleep 2
done
hosts_known=$(wc -l hosts_known.txt)
echo "We have $hosts_known known hosts that could be still active via SSH
keys"
# TIME TO TEST WHICH SSH servers are still active with the SSH keys
# AND THIS IS WHERE I AM FROZEN...
# Would love to have bash script that could
# ssh -l $users_that_have_/bin/bash -i $ssh_keys $ssh_servers
# Would also be very nice if it could save active
# SSH servers with the valid keys in output.txt in the format
# username:local-IP:/path/to/SSH_key
Please feel very comfortable to edit/modify the bash script above if it can serve better the goals described.
Any help would be very appreciated,
Thanks
The following works cool:
</etc/passwd \
grep /bin/bash |
cut -d: -f6 |
sudo xargs -i -- sh -c '
[ -e "$1" ] && cat "$1"
' -- {}/.ssh/known_hosts |
cut -d' ' -f1 |
tr ',' '\n' |
sed '
/^\[/{
s/\[\(.*\)\]:\(.*\)/\1 \2/;
t;
};
s/$/ 22/;
' |
sort -u |
xargs -l1 -- sh -c '
if echo "~" | nc -q1 -w3 "$1" "$2" | grep -q "^SSH"; then
echo "#### SUCCESS $1 $2";
else
echo "#### ERROR $1 $2";
fi
' --
So:
Start with /etc/passwd
Filter all "bash_users" as you call them
Filter user home directories only cut -d: -f6
For each user home directory sudo xargs -i -- run
Check if the file .ssh/known_hosts inside the user home directory exists
If it does, print it
Filter only hosts names
Multiple hosts signatures may share same key and are separated by a comma. Replace comma for newline
Now a sed script:
If a line starts with a [ that means it has a format of [host]:port and I want to replace it with host port
If the line does not start with a [ I add 22 to the end of the line so it's host 22
Then I sort -u
Now for each line:
I get the ssh version from ssh echo "~" | nc hostname port returns smth like "SSH-2.0-OpenSSH_6.0" + newline + "Protocol mismatch".
So if the line returned by nc hostname port starts with SSH that means there is ssh running on the other side
I added timeout for unresponsive hosts, but I think nc -w timeout option may also be used. Probably also nc -q 1 should be specified.
Now the real fun is, when you add the max-procs option to the last xargs line, you can check all hosts simultaneously. On my host I have 47 unique addresses and xargs -P30 checks them ALL in like 2 seconds.
But really there are some problems. The script needs root to read from all users known_hosts. But worse, the known_hosts may be hashed. It would be better to firstly know the list of hosts on your network, and then generate known_hosts from it. It would look like ssh-keyscan -f list_of_hosts > ~/.ssh/known_hosts or similar. Generaly ssh-keygen -F hostname should be used if a host exists in known_hosts, sadly there is no listing command. known_hosts file format may be found in ssh documentation.
I want a script which will restart the defined services on their respective servers. I want to pass parameter as below to the script:
eg:
sh execute.sh
Enter your server and service list: [server1:nginx,mysql],[server2:mysql,apache],[server3:mongodb,apache]
By this input the script should verify and start the services on the respective servers. I am able to do this on a single server by declaring variables.
#!/bin/bash
Instance_Name=server1
Service_Name=(nginx php-fpm mysql)
SSH_USER=admin
SSH_IDENT_FILE=~/credentials/user.pem
len=${#Service_Name[*]}
i=0
while [ $i -lt $len ]; do
service=${Service_Name[$i]}
ssh -i $SSH_IDENT_FILE -o StrictHostKeyChecking=no $SSH_USER#$Instance_Name 'service $service restart'
done
Now I don't have an idea to move forward. Please let me know if my question is unclear. Thanks in advance.
Parse your input parameters
# create an array of server:services
a=($(echo "$1" | gawk 'BEGIN { FS="[]],[[]" } ; { print $1, $2, $3 }' | tr -d '[]'))
# add a for loop here to iterate values in array with code below
for var in "${a[#]}" ; do
# get server name
server1=$(echo $var | cut -d ':' -f1)
# get your services as space separated
servs1="$(echo $var | cut -d ':' -f2 | tr ',' ' ')"
# loop your services
for s in $servs1; do
ssh $server1 "service $s restart"
done
done
If you like bash programming or have to learn it this is the 'bible' to me
Advanced Bash-Scripting Guide
An in-depth exploration of the art of shell scripting
Mendel Cooper
http://www.tldp.org/LDP/abs/html/
Try this?
#!/bin/bash
SSH_IDENT_FILE=~/credentials/user.pem
SSH_USER=admin
# tells 'for' to read "per line"
IFS='
'
for line in $(
# read input from 1st command-line arg (not stdin)
echo "$1"|\
# remove all services except (nginx php-fpm mysql)
sed 's/\([:,]\)\(\(nginx\|php-fpm\|mysql\)\|[^]:,[]\+\)/\1\3/g'|\
# make it multi-line:
# server1 svc1 svc2
# server2 svc1
sed 's/\[[^]:]*:,*\]//g;s/\],*$//;s/\],*/\n/g;s/[:,]\+/ /g;s/\[//g'|\
# make it single-param:
# server1 svc1
# server1 svc2
# server2 svc1
awk '{for(i=2;i<=NF;i++)print $1" "$i}'
);do
IFS=' ' read Instance_Name service <<< $line
echo ssh -i $SSH_IDENT_FILE -o StrictHostKeyChecking=no $SSH_USER#$Instance_Name "service $service restart"
done
How can I write a Linux Bash script that tells me which computers are ON in my LAN?
It would help if I could give it a range of IP addresses as input.
I would suggest using nmap's ping-scan flag,
$ nmap -sn 192.168.1.60-70
Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2009-04-09 20:13 BST
Host machine1.home (192.168.1.64) appears to be up.
Host machine2.home (192.168.1.65) appears to be up.
Nmap finished: 11 IP addresses (2 hosts up) scanned in 0.235 seconds
That said, if you want to write it yourself (which is fair enough), this is how I would do it:
for ip in 192.168.1.{1..10}; do ping -c 1 -t 1 $ip > /dev/null && echo "${ip} is up"; done
..and an explanation of each bit of the above command:
Generating list of IP addresses
You can use the {1..10} syntax to generate a list of numbers, for example..
$ echo {1..10}
1 2 3 4 5 6 7 8 9 10
(it's also useful for things like mkdir {dir1,dir2}/{sub1,sub2} - which makes dir1 and dir2, each containing sub1 and sub2)
So, to generate a list of IP's, we'd do something like
$ echo 192.168.1.{1..10}
192.168.1.1 192.168.1.2 [...] 192.168.1.10
Loops
To loop over something in bash, you use for:
$ for thingy in 1 2 3; do echo $thingy; done
1
2
3
Pinging
Next, to ping.. The ping command varies a bit with different operating-systems, different distributions/versions (I'm using OS X currently)
By default (again, on the OS X version of ping) it will ping until interrupted, which isn't going to work for this, so ping -c 1 will only try sending one packet, which should be enough to determine if a machine is up.
Another problem is the timeout value, which seems to be 11 seconds on this version of ping.. It's changed using the -t flag. One second should be enough to see if a machine on the local network is alive or not.
So, the ping command we'll use is..
$ ping -c 1 -t 1 192.168.1.1
PING 192.168.1.1 (192.168.1.1): 56 data bytes
--- 192.168.1.1 ping statistics ---
1 packets transmitted, 0 packets received, 100% packet loss
Checking ping result
Next, we need to know if the machine replied or not..
We can use the && operator to run a command if the first succeeds, for example:
$ echo && echo "It works"
It works
$ nonexistantcommand && echo "This should not echo"
-bash: nonexistantcommand: command not found
Good, so we can do..
ping -c 1 -t 1 192.168.1.1 && echo "192.168.1.1 is up!"
The other way would be to use the exit code from ping.. The ping command will exit with exit-code 0 (success) if it worked, and a non-zero code if it failed. In bash you get the last commands exit code with the variable $?
So, to check if the command worked, we'd do..
ping -c 1 -t 1 192.168.1.1;
if [ $? -eq 0 ]; then
echo "192.168.1.1 is up";
else
echo "ip is down";
fi
Hiding ping output
Last thing, we don't need to see the ping output, so we can redirect stdout to /dev/null with the > redirection, for example:
$ ping -c 1 -t 1 192.168.1.1 > /dev/null && echo "IP is up"
IP is up
And to redirect stderr (to discard the ping: sendto: Host is down messages), you use 2> - for example:
$ errorcausingcommand
-bash: errorcausingcommand: command not found
$ errorcausingcommand 2> /dev/null
$
The script
So, to combine all that..
for ip in 192.168.1.{1..10}; do # for loop and the {} operator
ping -c 1 -t 1 192.168.1.1 > /dev/null 2> /dev/null # ping and discard output
if [ $? -eq 0 ]; then # check the exit code
echo "${ip} is up" # display the output
# you could send this to a log file by using the >>pinglog.txt redirect
else
echo "${ip} is down"
fi
done
Or, using the && method, in a one-liner:
for ip in 192.168.1.{1..10}; do ping -c 1 -t 1 $ip > /dev/null && echo "${ip} is up"; done
Problem
It's slow.. Each ping command takes about 1 second (since we set the -t timeout flag to 1 second). It can only run one ping command at a time.. The obvious way around this is to use threads, so you can run concurrent commands, but that's beyond what you should use bash for..
"Python threads - a first example" explains how to use the Python threading module to write a multi-threaded ping'er.. Although at that point, I would once again suggest using nmap -sn..
In the real world, you could use nmap to get what you want.
nmap -sn 10.1.1.1-255
This will ping all the addresses in the range 10.1.1.1 to 10.1.1.255 and let you know which ones answer.
Of course, if you in fact want to do this as a bash exercise, you could run ping for each address and parse the output, but that's a whole other story.
Assuming my network is 10.10.0.0/24, if i run a ping on the broadcast address like
ping -b 10.10.0.255
I'll get an answer from all computers on this network that did not block their ICMP ping port.
64 bytes from 10.10.0.6: icmp_seq=1 ttl=64 time=0.000 ms
64 bytes from 10.10.0.12: icmp_seq=1 ttl=64 time=0.000 ms
64 bytes from 10.10.0.71: icmp_seq=1 ttl=255 time=0.000 ms
So you just have to extract the 4th column, with awk for example:
ping -b 10.10.0.255 | grep 'bytes from' | awk '{ print $4 }'
10.10.0.12:
10.10.0.6:
10.10.0.71:
10.10.0.95:
Well, you will get duplicate, and you may need to remove the ':'.
EDIT from comments :
the -c option limits the number of pings
since the script will end, we can also limit ourself on unique IPs
ping -c 5 -b 10.10.0.255 | grep 'bytes from' | awk '{ print $4 }' | sort | uniq
There is also fping:
fping -g 192.168.1.0/24
or:
fping -g 192.168.1.0 192.168.1.255
or show only hosts that are alive:
fping -ag 192.168.1.0/24
It pings hosts in parallel so the scan is very fast. I don't know a distribution which includes fping in its default installation but in most distributions you can get it through the package manager.
Also using the "ping the broadcast address" method pointed out by chburd, this pipe should do the trick for you:
ping -c 5 -b 10.11.255.255 | sed -n 's/.* \([0-9]\+\.[0-9]\+\.[0-9]\+\.[0-9]\+\).*/\1/p' | sort | uniq
Of course, you'd have to change the broadcast address to that of your network.
Just for fun, here's an alternate
#!/bin/bash
nmap -sP 192.168.1.0/24 > /dev/null 2>&1 && arp -an | grep -v incomplete | awk '{print$2}' | sed -e s,\(,, | sed -e s,\),,
If you're limiting yourself to only having the last octet changing, this script should do it. It should be fairly obvious how to extend it from one to multiple octets.
#! /bin/bash
BASE=$1
START=$2
END=$3
counter=$START
while [ $counter -le $END ]
do
ip=$BASE.$counter
if ping -qc 2 $ip
then
echo "$ip responds"
fi
counter=$(( $counter + 1 ))
done
ip neighbor
arp -a
Arpwatch
As other posters pointed out, nmap is the way to go, but here's how to do the equivalent of a ping scan in bash. I wouldn't use the broadcast ping, as a lot of systems are configured not to respond to broadcast ICMP nowadays.
for i in $(seq 1 254); do
host="192.168.100.$i"
ping -c 1 -W 1 $host &> /dev/null
echo -n "Host $host is "
test $? -eq 0 && echo "up" || echo "down"
done
#!/bin/bash
#Get the ip address for the range
ip=$(/sbin/ifconfig eth0 | grep 'inet addr:' | cut -d: -f2 | awk '{ print $1}' | cut -d"." -f1,2,3)
# ping test and list the hosts and echo the info
for range in $ip ; do [ $? -eq 0 ] && ping -c 1 -w 1 $range > /dev/null 2> /dev/null && echo "Node $range is up"
done
Although an old question, it still seems to be important (at least important enough for me to deal with this). My script relies on nmap too, so nothing special here except that ou can define which interface you want to scan and the IP Range is created automagically (at least kind of).
This is what I came up with
#!/bin/bash
#Script for scanning the (local) network for other computers
command -v nmap >/dev/null 2>&1 || { echo "I require nmap but it's not installed. Aborting." >&2; exit 1; }
if [ -n ""$#"" ]; then
ip=$(/sbin/ifconfig $1 | grep 'inet ' | awk '{ print $2}' | cut -d"." -f1,2,3 )
nmap -sP $ip.1-255
else
echo -e "\nThis is a script for scanning the (local) network for other computers.\n"
echo "Enter Interface as parameter like this:"
echo -e "\t./scannetwork.sh $(ifconfig -lu | awk '{print $2}')\n"
echo "Possible interfaces which are up are: "
for i in $(ifconfig -lu)
do
echo -e "\033[32m \t $i \033[39;49m"
done
echo "Interfaces which could be used but are down at the moment: "
for i in $(ifconfig -ld)
do
echo -e "\033[31m \t $i \033[39;49m"
done
echo
fi
One remark: This script is created on OSX, so there might be some changes to linux environments.
If you want to provide a list of hosts it can be done with nmap, grep and awk.
Install nmap:
$ sudo apt-get install nmap
Create file hostcheck.sh like this:
hostcheck.sh
#!/bin/bash
nmap -sP -iL hostlist -oG pingscan > /dev/null
grep Up pingscan | awk '{print $2}' > uplist
grep Down pingscan | awk '{print $2}' > downlist
-sP: Ping Scan - go no further than determining if host is online
-iL : Input from list of hosts/networks
-oG : Output scan results in Grepable format, to the given filename.
/dev/null : Discards output
Change the access permission:
$ chmod 775 hostcheck.sh
Create file hostlist with the list of hosts to be checked (hostname or IP):
hostlist (Example)
192.168.1.1-5
192.168.1.101
192.168.1.123
192.168.1.1-5 is a range of IPs
Run the script:
./hostcheck.sh hostfile
Will be generated files pingscan with all the information, uplist with the hosts online (Up) and downlist with the hosts offline (Down).
uplist (Example)
192.168.1.1
192.168.1.2
192.168.1.3
192.168.1.4
192.168.1.101
downlist (Example)
192.168.1.5
192.168.1.123
Some machines don't answer pings (e.g. firewalls).
If you only want the local network you can use this command:
(for n in $(seq 1 254);do sudo arping -c1 10.0.0.$n & done ; wait) | grep reply | grep --color -E '([0-9]+\.){3}[0-9]+'
Explanations part !
arping is a command that sends ARP requests. It is present on most of linux.
Example:
sudo arping -c1 10.0.0.14
the sudo is not necessary if you are root ofc.
10.0.0.14 : the ip you want to test
-c1 : send only one request.
&: the 'I-don't-want-to-wait' character
This is a really useful character that give you the possibility to launch a command in a sub-process without waiting him to finish (like a thread)
the for loop is here to arping all 255 ip addresses. It uses the seq command to list all numbers.
wait: after we launched our requests we want to see if there are some replies. To do so we just put wait after the loop.
wait looks like the function join() in other languages.
(): parenthesis are here to interpret all outputs as text so we can give it to grep
grep: we only want to see replies. the second grep is just here to highlight IPs.
hth
Edit 20150417: Maxi Update !
The bad part of my solution is that it print all results at the end. It is because grep have a big enough buffer to put some lines inside.
the solution is to add --line-buffered to the first grep.
like so:
(for n in $(seq 1 254);do sudo arping -c1 10.0.0.$n & done ; wait) | grep --line-buffered reply | grep --color -E '([0-9]+\.){3}[0-9]+'
#!/bin/bash
for ((n=0 ; n < 30 ; n+=1))
do
ip=10.1.1.$n
if ping -c 1 -w 1 $ip > /dev/null 2> /dev/null >> /etc/logping.txt; then
echo "${ip} is up" # output up
# sintax >> /etc/logping.txt log with .txt format
else
echo "${ip} is down" # output down
fi
done
The following (evil) code runs more than TWICE as fast as the nmap method
for i in {1..254} ;do (ping 192.168.1.$i -c 1 -w 5 >/dev/null && echo "192.168.1.$i" &) ;done
takes around 10 seconds, where the standard nmap
nmap -sP 192.168.1.1-254
takes 25 seconds...
Well, this is part of a script of mine.
ship.sh 🚢 A simple, handy network addressing 🔎 multitool with plenty of features 🌊
Pings network, displays online hosts on that network with their local IP and MAC address
It doesn't require any edit. Needs root permission to run.
GOOGLE_DNS="8.8.8.8"
ONLINE_INTERFACE=$(ip route get "${GOOGLE_DNS}" | awk -F 'dev ' 'NR == 1 {split($2, a, " "); print a[1]}')
NETWORK_IP=$(ip route | awk "/${ONLINE_INTERFACE}/ && /src/ {print \$1}" | cut --fields=1 --delimiter="/")
NETWORK_IP_CIDR=$(ip route | awk "/${ONLINE_INTERFACE}/ && /src/ {print \$1}")
FILTERED_IP=$(echo "${NETWORK_IP}" | awk 'BEGIN{FS=OFS="."} NF--')
ip -statistics neighbour flush all &>/dev/null
echo -ne "Pinging ${NETWORK_IP_CIDR}, please wait ..."
for HOST in {1..254}; do
ping "${FILTERED_IP}.${HOST}" -c 1 -w 10 &>/dev/null &
done
for JOB in $(jobs -p); do wait "${JOB}"; done
ip neighbour | \
awk 'tolower($0) ~ /reachable|stale|delay|probe/{printf ("%5s\t%s\n", $1, $5)}' | \
sort --version-sort --unique