High Performance Computing Cluster Setup Manuelx CHPC Guide

User Manual: Pdf

Open the PDF directly: View PDF .
Page Count: 17

Download
Open PDF In Browser	View PDF

High Performance Computing Cluster Setup Manuel
OS
Convert SDD to non-RAID!!!
When you go into BIOS to change boot order to install in OS, also set the BIOS to performance mode
Network Setup Initial
Type “nmcli d” command in your terminal for quick list of ethernet cards installed on your machine
Type “nmtui” command in your terminal to open Network manager. After opening Network manager
chose “Edit connection” and press Enter (Use TAB button for choosing options).
Now choose you network interfaces and click “Edit”
Choose “Automatic” in IPv4 CONFIGURATION and check Automatically connect check box and press OK
and quit from Network manager.
Restart network services: service network restart
Make sure nmtui is done by now and never do it again. It changes the network scripts
/etc/sysconfig/network-scripts for the files for the ethernet ports
Edit the network scripts on all the nodes, for ifcfg-enp8s0f0 and ifcfg-enp8s0f1, found in:
/etc/sysconfig/network-scripts
Proxy (for wits, obviously the details will change depending on location):
The following lines should appear in the .bashrc file (these are for wits)
export
http_proxy='http://students\12345:password@proxyss.wits.ac.za:80’ in
.bash_profile
export
https_proxy='https://students\12345:password@proxyss.wits.ac.za:80’ in
.bash_profile
export
HTTP_PROXY='http://students\12345:password@proxyss.wits.ac.za:80’ in
.bash_profile
export
HTTPS_PROXY='https://students\12345:password@proxyss.wits.ac.za:80’ in
.bash_profile
Craig’s guide:

export HTTP_PROXY=$http_proxy
export HTTPS_PROXY=$https_proxy

source .bashrc

Stop the firewall on all the nodes:
systemctl disable firewalld
systemctl stop firewalld
systemctl status firewalld
Adds a static route:
ip route add destination_node_connected_port dev port_connected_to_that_node_from_local_node
Eg:
ip route add 10.0.0.92 dev enp8s0f0
To add a gateway (CIDR Notation):
ip route add category_of_IP_addresses_to_use_gateway via
IP_of_port_on_connected_node_that_local_node_is_sending_traffic_through
Eg:
ip route add 10.0.0.0/24 via 10.0.0.92 (10.0.0.93 being source node ip sending
traffic through 10.0.0.92)
systemctl restart network
Eg for beyonce to kelly:
ip route add 10.0.0.91 dev enp8s0f1
Eg for kelly to beyonce:
ip route add 10.0.0.90 dev enp8s0f0
ip route add 10.0.0.0/24 via 10.0.0.90
Eg for Kelly to michelle
ip route add 10.0.0.93 dev enp8s0f1
Eg for michelle to Kelly:
ip route add 10.0.0.92 dev enp8s0f0
ip route add 10.0.0.0/24 via 10.0.0.92
ip route add 10.0.0.90 via 10.0.0.92
Eg beyonce to michelle:
ip route add 10.0.0.93 via 10.0.0.91
We need to setup the following:
-beyonce connect to Kelly
-kelly connect to beyonce
-kelly connect to michelle
-michelle connect to Kelly
-beyonce special gateway to michelle
-michelle special gateway to beyonce
---Explanation of gateway ip rule:

Destination is where we are going, gateway is how you are going to get there, so we'd have to
go through kelly to get to michelle from beyonce. You must add all special ip addresses into the
iptables.
---Explanation of /24 in gateway command:
The /24 indicates the amount of bits variance that is allowed in the IP addresses for them to still
use this gateway. So, in this case the first 24 bits are fixed while the last 8 bits may vary
(32 bits in an IP). Visually: xxxx.xxxx.xxxx.0000 where x’s are fixed and 0’s may vary.
IP addresses are broken up into 4 parts of 8 bits representing integers between 0 and 255
(highest number 8 bits can represent). The netmask can be interpreted as the amount of
restrictions placed on the IP address. So, 255.255.255.255 means only 1 address can be used, or
only one address will pass through that gateway. 0.0.0.0 places no restrictions and thus any
range of IP’s will pass through the gateway. In finality, a gateway of 10.0.0.0/24 will support
a range of IP addresses from 10.0.0.1 to 10.0.0.255 and the netmask should be
255.255.255.0
To delete a gateway:
ip route del connection
Eg:
ip route del 10.0.0.93
IP Tables and NAT

To show routing table: ip route OR route -n (net-tools needs to be installed)
Make sure firewalld isn’t running:
systemctl disable firewalld
systemctl stop firewalld
systemctl status firewalld
Flush IP-tables:
iptables --flush
iptables -–flush -t nat
sysctl -w net.ipv4.ip_forward=1
vi /etc/sysctl.conf and add (only in nodes that need to forward incoming connections):
net.ipv4.ip_forward = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.all.disable_ipv6 = 1
sysctl -p /etc/sysctl.conf
(Enables the changes)
vi /etc/resolv.conf and make sure it says:
search ms.wits.ac.za
nameserver 146.141.8.16
nameserver 146.141.21.2
nameserver 146.141.15.210
nameserver 146.141.15.222

scp /etc/resolv.conf root@node1:/etc/ (This is the DNS server list that was
generated on headnode that needs
to be copied to compute nodes
so that they can access servers
by their names)
This will probably change if you’re not at wits. It should be generated by dhclient-script
To forward data through nodes (only to be done in nodes actually forwarding data):
iptables -t nat -A POSTROUTING -o port_of_outgoing_connection_to_forward_through -j MASQUERADE
Eg:
iptables -t nat -A POSTROUTING -o enp8s0f0 -j MASQUERADE
iptables-save
To allow through firewall- only if firewall is running:
iptables -A FORWARD -i port_of_incoming_connection_to_forward_from -n state
Eg:
iptables -A FORWARD -i enp8s0f0 (compute node) -n state
iptables --flush
iptables -t nat –flush
iptables -L -t nat
iptables -L
iptables-save

to clear ip tables
to clear nat ip tables
to view nat routing tables
to view normal routing tables
saves routing tables

Copy /etc/resolv.conf file from headnode into compute nodes (Instead of doing dns):
scp /etc/resolv.conf root@compute_nodes:/etc/

If you are restarting one node:
There will be a problem with ssh’ing into the restarted node. The solution is to delete the known hosts
entry for the restarted node in the two other nodes:
vi .ssh/known_hosts and delete both the nickname entry and actual ip address entry for the
restarted node
Changing Hostname:
hostname "name" Eg: hostname beyonce (without inverted commas)
vi /etc/sysconfig/network and add HOSTNAME="name" entry (without inverted commas)
vi /etc/hostname and type hostname (actual name of node, eg: beyonce in beyonce) by itself and
remove the localhost.localdomain
vi /etc/hosts and add all hosts and all corresponding ports of this cluster into this file (internal and
external ports. Some names will be registered to a few ports)
Adding to the /etc/hosts file:
10.0.0.90 beyonce

10.0.0.91 kelly
10.0.0.92 kelly
10.0.0.93 michelle
To setup passwordless ssh (repeat for every node moving the key to all other nodes):
1)Generate key: ssh-keygen -t rsa
2) ssh-copy-id -i ~/.ssh/id_rsa.pub root@"ip-address" to copy the key to the
other nodes. Thus, you should generate 3 keys and copy 6 times for a 3-node cluster.

NFS:
yum install nfs-utils on all nodes
setsebool _p use_nfs_home_dirs 1 (This gives permissions for using nfs folder (run on
startup every time))

systemctl enable and start (1)
rpcbind
systemctl enable and start (2)
nfs-server
systemctl enable and start (3)
nfs-lock
systemctl enable and start (4)
nfs-idmap
Make a directory for mounting (probably in root, call nfsmount): mkdir /nfsmount
vim /etc/exports and add the directory for your mount (only on headnode), so add
/nfsmount:
/nfsmount *(rw,async,insecure,no_root_squash)
systemctl restart nfs-server

chmod 777 /nfsmount (In headnode give everyone permissions for the directory)
Then go to compute node:

And re-run (1) to (4) on compute node

Make directory to mount (probably also /nfsmount)
vim /etc/fstab and add the line:

headnode:/nfsmount /nfsmount nfs
rw,sync,hard,intr
(headnode:/nfsmount is directory of mounted file in head)
(/nfsmount is the directory of the mounted folder on the compute node)

(use tabs between parts except between the two 0’s)

0 0

Restart nfs-server in headnode:
systemctl restart nfs-server
mount -a (mounts everything in /etc/fstab (all the directories that are mounted))
df (lists all mounted devices).
NFS essentially makes the mounted directory act as a hard drive on the other node (works both
ways. So, the mounted directory becomes common on both nodes)
If you get a stale mount:
systemctl restart nfs-server #on headnode
systemctl restart nfs #on headnode
mount -a
#on compute nodes
Mounting an NTFS Drive:
dmesg
(Shows the connected usb devices at the end of its printout)
yum install epel-release (Install EPEL) (Must be installed before ntfs-3g)
yum install ntfs-3g
(Install the ntfs-3g driver package)
mkdir /mnt/win
(create a directory where the NTFS drive shall be mounted)
df -h
(Shows connected devices)
mount -t ntfs-3g /dev/sdb1 /mnt/win
(mount the NTFS partition)
Note: In this example the NTFS partition is the device /dev/sdb1, you have to replace that with the
device name of your NTFS partition.
The mount point will exist until reboot or until you unmount it
umount /mnt/win

(Unmount drive, specifies directory where the device is mounted)

To mount the NTFS partition permanently:
vim /etc/fstab
Add the following line to fstab:
/dev/sdb1 /mnt/win ntfs-3g defaults 0 0
Again changing /dev/sdb1 to the name of your device on the system

Mounting a USB:
dmesg
mount /dev/sdb1 /mnt
fdisk -l
umount /mnt

(To see list of connected devices)
(Change /dev/sdb1 to the name of your device on your system)
(unmount a usb)

Creating Users (While as root user)
adduser username
passwd username

(You’ll be prompted to enter the password twice)

usermod -u UID username

gpasswd -a username wheel
sudo lid -g wheel
userdel username
userdel -r username
su username

(To change user UID, they must correlate on all nodes per user,
UID is where you enter the value, 0 to 99 is reserved for the
system)
(To grant sudo privileges)
(To show a list of users with sudo privileges)
(To delete a user)
(To delete a user along with all their files)
(switch users)

Installing Intel Compiler (only on headnode, share to compute nodes by nfs sharing /opt):
Download from http://registrationcenterdownload.intel.com/akdlm/irc_nas/tec/12374/parallel_studio_xe_2018_update1_cluster_edition.tgz
Serial number:
cd Intel\ compiler/
tar -xvzf parallel_studio_xe_2018_cluster_edition.tgz
cd parallel_studio_xe_2018_cluster_edition
./install.sh (Will install in /opt if you use the default)
source parallel_studio_xe_2018.0.033/psxevars.sh (add into .bashrc as well, with its
full file path in opt)
mpicc
Setting up compiler after installation
In compute nodes and on headnode install the following:
yum install gcc5 gcc gtk kernel-devel kernel-PAE-devel libgtk-3-dev
libgtk gtk+-devel gtk2-devel wget
vim .bashrc
Add the following to bashrc if it’s hasn’t already been added:
source /opt/intel/parallel_studio_xe_2018.0.033/psxevars.sh
#Run hpcc-1.5.0/hpl
HPCC
Helpful Links:
http://wiki.chpc.ac.za/acelab:hpcc

CHPC installation guide

http://icl.cs.utk.edu/hpcc/
HPCC Website
http://hpl-calculator.sourceforge.net/
Theoretical Peak Performance Calculator
https://software.intel.com/en-us/articles/performance-tools-for-software-developers-hplapplication-note
THIS ONE IS GOOD!
Downloading the source:
cd ~
mkdir HPCC
cd HPCC
wget http://icl.cs.utk.edu/projectsfiles/hpcc/download/hpcc1.5.0a.tar.gz
tar -xf hpcc-1.5.0a.tar.gz
cd hpcc-1.5.0a
cd HPCC/hpcc-1.5.0a/hpl/setup
Use the Makefile which is adapted from the HPL which ships with the Intel MKL package:
cp Makefile.LinuxIntelIA64Itan2_eccMKL ~/HPCC/hpcc-1.5.0a/hpl
#cp Makefile.intel64 ~/HPCC/hpcc-1.5.0a/hpl
Make with:
make arch=LinuxIntelIA64Itan2_eccMKL
#make arch=intel64
mv Makefile.LinuxIntelIA64Itan2_eccMKL Makefile.destiny (rename make file)
vim Makefile.destiny
Don’t edit anything until LAdir
Then make the following changes:
Point the math library MKL:
LAdir = = /opt/intel/compilers_and_libraries_2018/linux/mkl
LAinc = = -I$(LAdir)/include
LAlib = = -L$(LAdir)/lib/intel64 -mkl

Change the compiler information to use intel compiler:
CC
CCNOOPT

= mpiicc
= $(HPL_DEFS)

CCFLAGS
= = $(HPL_DEFS) -std=c99 -fomit-frame-pointer -O3 funroll-loops -W -Wall (With an o, the capital letter, in O3)
LINKER
LINKFLAGS

= mpiicc
= $(CCFLAGS)

mv _hpccinf.txt hpccinf.txt (Remove underscore in front)
make arch=destiny clean
make arch=destiny
NFS mount the hpcc-1.5.0 directory across the whole cluster
Benchmarking:
Before running the benchmark, edit the hpccinf.txt file. Set appropriate N, NB, P and Q values
(using the calculator given above try get as close to peak performance as possible)
Run the benchmark using:
mpirun -np  -hostfile  ./hpcc
Where  is the number of cores available to the system and  is your hostsfile
Eg:
mpirun -np 4 hpcc
mpirun -np 24 -hosts node0,node1,node2 ./hpcc to run the
full benchmark
Add hostnames to the hostfile:
In hpcc directory (probably in /nfsmount)
vim hostfile
add name of hosts (beyonce, kelly, michelle) on their own line underneath each other
Formatting the output:
Download the tar file to format the output from the link:
http://wiki.chpc.ac.za/lib/exe/fetch.php?tok=8dd21b&media=http%3A%2F%2Fwiki.chpc.ac.za%
2F_media%2Facelab%3Aformat.tar
Run:
./format.pl -w -f hpccoutf.txt
To give the output from hpccoutf.txt (the results from the benchmark)in the correct
format
HPCC Key:
N=problem size
Ps
Qs
NB = determine block size
P and Q close together (try make the matrix square)
If you get RLIMIT MEMLOCK TOO SMALL:
ulimit -l unlimited
vim /etc/security/limits.conf:
add the following two lines

*
soft memlock
*
hard memlock
Using tabs in the commands above

unlimited
unlimited

Helpful commands:
tail -n 200 “filename”| less

to see output from files

To check if the benchmark is running on all nodes:
run the benchmark in the background by using & at the end of the command to run it
ssh into each node
top (the process should be running in each node)

WRF (Put in home, not in root) (Try option 15 instead of option 20):
Useful Links:
http://www2.mmm.ucar.edu/wrf/OnLineTutorial/compilation_tutorial.php
http://www.hpcadvisorycouncil.com/pdf/WRF_v3.8%20Installation_Best_Practices.pdf
http://wiki.chpc.ac.za/acelab:wrf
(CHPC Guide)
The second link includes how to compile with pnetcdf (which may be slower. Using pnetcdf
instead of netcdf may or may not be useful. If you have time compile both and compare
performance with both and choose whichever performs better.)
Additional packages need to be installed:
yum install m4 csh perl perl5 –y
Add the following environment variables to .bashrc for WRF:
vim ~/.bashrc
export CC=icc
export
export
export
export
export
export
export
export
export
export
export
export
export
export
export
export
export
export

CXX=icpc
CFLAGS='-O3 -xHost -ip -no-prec-div -static-intel'
CXXFLAGS='-O3 -xHost -ip -no-prec-div -static-intel'
F77=ifort
FC=ifort
F90=ifort
FFLAGS='-O3 -xHost -ip -no-prec-div -static-intel'
CPP='icc -E'
CXXCPP='icpc -E'
DIR=~/WRF/libs
PATH=$DIR/netcdf/bin:$PATH
NETCDF=$DIR/netcdf
LDFLAGS=-L$DIR/grib2/lib
LD_LIBRARY_PATH=$DIR/grib2/lib:$LD_LIBRARY_PATH
CPPFLAGS=-I$DIR/grib2/include
JASPERLIB=$DIR/grib2/lib
JASPERINC=$DIR/grib2/include
WRFIO_NCD_LARGE_FILE_SUPPORT=1

export PATH=~/WRF/WPS:$PATH
export PATH=~/WRF/WRFV3/run:$PATH

export DM_FC=mpiifort
export DM_CC=mpiicc
Make sure the exported path or directory in .bashrc is the correct directory of the WRF libs:
Export DIR=/home/$USER/WRF/libs
for example, should be used if WRF is in /root/ this could be different, so you need to check
it in relation to the current system
make sure we have libstdc++, gfortran and gcc (can check by running which gcc):
yum install libstdc++ gfortran gcc
yum install gcc-c++ zlib libpng
Set up the directory:
mkdir WRF
cd WRF
mkdir libs
mkdir tars
vim .bashrc and make sure that source
parallel_studio_xe_2018.0.033/psxevars.sh appears in it
source .bashrc
Jasper Setup:
cd ~/WRF/tars
wget
http://www2.mmm.ucar.edu/wrf/OnLineTutorial/compile_tutorial/tar_
files/jasper-1.900.1.tar.gz
tar -xf jasper-1.900.1.tar.gz
cd jasper-1.900.1
./configure --prefix=/home/$USER/WRF/libs/grib2
make -j20
make -j20 install

NetCDF Setup:

cd ~/WRF/tars
wget
http://www2.mmm.ucar.edu/wrf/OnLineTutorial/compile_tutorial/tar_
files/netcdf-4.1.3.tar.gz

tar -xf netcdf-4.1.3.tar.gz
cd netcdf-4.1.3
./configure --prefix=/home/$USER/WRF/libs/netcdf --disable-dap
--disable-netcdf-4 --disable-shared (May have to remove shared flag)
make -j20

(Compiles on all 12 cores, number changes depending on number of cores)

make -j20 check
make -j20 install
WRF Setup:
cd ~/WRF/tars
wget http://www2.mmm.ucar.edu/wrf/src/WRFV3.7.TAR.gz
tar -xf WRFV3.7.TAR.gz
mv WRFV3 ..
cd ../WRFV3
./configure
You will be prompted to select config options for WRF. NetCDF should be found from
environment variables set above:
for Linux x86_64 options: select 20 (may be 15, hardware dependent),
then option 1 for compile nesting
After the configure step is complete, build the code with:
./compile -j20 em_real(space or no space it does not matter
for j20)
WPS Setup:
cd ~/WRF/tars
wget http://www2.mmm.ucar.edu/wrf/src/WPSV3.7.TAR.gz
tar -xf WPSV3.7.TAR.gz
mv WPS ..
cd ../WPS
./configure
Select option 17: Linux, with Intel, serial /w grib2
./compile

ln -sf ungrib/Variable_Tables/Vtable.GFS Vtable

Benchmarking:
cd ~/WRF/tars
Copy the input data from:
http://www.ace.chpc.ac.za/tars/GEOG.tar.gz
wget http://www.ace.chpc.ac.za/tars/GEOG.tar.gz
tar -xf GEOG.tar.gz
mv GEOG ..
cd ../GEOG
cp namelist.wps ../WPS
cd ~/WRF/tars
Copy more input data from
http://www.ace.chpc.ac.za/tars/DATA.tar.gz
wget http://www.ace.chpc.ac.za/tars/GEOG.tar.gz
tar -xf DATA.tar.gz
mv DATA ..
cd ../WPS
./link_grib.csh ../DATA/

Build the model from the geographical data:
./geogrid.exe
./ungrib.exe
./metgrid.exe
Set up the benchmark:
cd ~/WRF
cp GEOG/namelist.input WRFV3/run
cd WRFV3/run

ln -sf ../../WPS/met_em.d01.2015* .

(With the full stop at the end)

Final setup step:
ulimit -s unlimited (makes stack size unlimited, stops segfaults)
If setting unlimited stack size above doesn’t fix segfaults, add the following to .bashrc:
(This is optional, don’t try initially with this)
vim ~/.bashrc
export OMP_STACKSIZE=64000000
It may not have made a difference but if problems persist then comment out all the
environment variables with “static” in them in .bashrc
./real.exe
Run the benchmark:
mpirun -np  -hostfile  ./wrf.exe
Where  = Number of cores to run the benchmark on
 = hostfile (can also use -hosts node0,node1,node2 in place of hostfile)
If the model segfaults at the very beginning of the run it is likely that there is a problem
with the input data.
If you run into problems try:
./clean -a
./configure
Choose 15 (as opposed to 20)
./compile em_real
Recompile WPS
If program doesn’t build properly:
cd WRFV3
vi configure.wrf and take out the word “time” in FC=
Note:

The dx and dy in WPS and WRFV3 have to correlate/match
Try removing all optimization flags and possibly using gcc instead of the Intel compiler if
all else fails.

LAMMPS
yum install libjpeg-turbo-devel (must be performed on all nodes)
Setup directory:

mkdir LAMMPS
cd LAMMPS
mkdir tars
vim .bashrc and make sure that source
parallel_studio_xe_2018.0.033/psxevars.sh appears in it
source ~/.bashrc
cd tar
Building the benchmark:
Download the LAMMPS source from: http://www.ace.chpc.ac.za/tars/lammps-stable.tar.gz
wget http://www.ace.chpc.ac.za/tars/lammps-stable.tar.gz
yum install libjpeg-turbo-devel (If it says we need a .h file we need the devel)
tar -xf lammps.stable.tar.gz
mv lammps-15May15 ..
cd ../lammps-15May15/
Edit the makefile:
cd src
cp MAKE/OPTIONS/Makefile.intel_cpu MAKE/Makefile.intel_cpu
vim MAKE/Makefile.intel_cpu
Edit FFT_INC to be:
FFT_INC =
-DFFT_MKL
Also edit LINKFLAGS and CCFLAGS, make openmp into qopenmp in the top
few lines.
For CCFLAGS, change -no-offload to -pno-offload and might need
to edit override to qoverride
vim MAKE/Makefile.intel_cpu
Set build parameters:
Make yes-opt
make yes-user-intel
make yes-user-omp
make intel_cpu
or use make intel_cpu -j12 to compile in parallel with 12 cores
The lmp_intel_cpu binary should be produced. (That’s an L in lmp)
cp lmp_intel_cpu ../bench
Add into .bashrc:
Export OMP_NUM_THREADS=12 (This may break the other benchmarks)
Benchmarking:

The stock “3d Lennard-Jones melt” test problem is used as a benchmark for this code. A
fixed number of particles/core is used for the problem size.
cd ~/LAMMPS/lammps-15May15/bench
Get the input script from
http://wiki.chpc.ac.za/lib/exe/fetch.php?tok=35559e&media=http%3A%2F%2Fwiki.chpc
.ac.za%2F_media%2Facelab%3Acpu.tar.gz
wget
http://wiki.chpc.ac.za/lib/exe/fetch.php?tok=35559e&media=h
ttp%3A%2F%2Fwiki.chpc.ac.za%2F_media%2Facelab%3Acpu.tar.gz
For x86 CPU benchmarks, this is 500K particles per core. Therefore, if you with to run
the benchmark on 24 x86 cores, a total of 12,000K (500×24) particles is required.
To run the benchmark use:
mpirun -np  -hostfile  ./lmp_intel_cpu -sf intel -v
x  -v y  -v z  -v t 100 < in.lj 2>&1 | tee
output.txt (have to submit output.txt)
where:  is the number of cores,  is the hostfile amd ,  and  are the
problem scaling factors - used to reach 500K particles/core. The benchmark has been
pre-configured to operate using 500K particles. Therefore, the X,Y and Z values are using
to scale the number of particles up - for more cores. In order to run the benchmark on
more cores simply scale X,Y and Z accordingly, such that their product equals the
number of cores desired.
For example, running on 4 nodes with 24 cores each, the run command would be:
mpirun -np 96 -hostfile hosts ./lmp_intel_cpu -sf
intel -v x 6 -v y 4 -v z 4 -v t 100 < in.lj
Giving a total particle count of 500K * (6*4*4) = 48×10^6. Which conforms to the 500K
particles per core, (48×10^6 / 96 = 500K).
If you get : “Error: rlimit memlock too small”:
in each node go to /etc/security/limits.conf and append
*
hard
memlock
unlimited
*
soft
memlock
unlimited
*
*

hard
soft

stack
stack

unlimited
unlimited

ulimit
ulimit
ulimit
ulimit
ulimit

-a
(Displays current limits)
-l unlimited
(That’s an L for -l)
-s unlimited
-l
(To check size of -l limit)
-s
(To check the size of the limit of the stack)

TMUX
start new:
tmux
start new session with name:
tmux new –s myname
attach:
tmux a
attach to named:
tmux a –t myname
list sessions:
tmux ls
kill session:
tmux kill-session –t myname
Create new windows:
Ctrl-b c
Shift to next window:
Ctrl-b n
Shift to previous window:
Ctrl-b p
List windows:
Ctrl-b w
Split screen vertically:
Ctrl-b %
Split screen horizontally:
Ctrl-b “
Make scrollable:
Ctrl-b [
Useful Commands:
For when we are copying or removing a file or directory and we do not wish to keep pressing “y”
Use: rm -r -f … (what we are trying to remove)
-r Flag = Directory
-f Flag = Force
To copy from directory to directory:
scp “file” “host”:”file”



---

Source Exif Data:

File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.7
Linearized                      : No
Warning                         : Info object (102 0 obj) not found at 681804

EXIF Metadata provided by EXIF.tools