update bugs: bos_image, basic_setup, shadow

Signed-off-by: tkucherera <[email protected]>

components/admin/docs/SPECS/docs.spec

@@ -103,6 +103,9 @@ make ; %{parser} steps.tex > recipe.sh ; popd
 pushd docs/recipes/install/rocky9/x86_64/xcat/slurm
 make ; %{parser} steps.tex > recipe.sh ; popd
 
+pushd docs/recipes/install/rocky9/x86_64/confluent/slurm
+make ; %{parser} steps.tex > recipe.sh ; popd
+
 pushd docs/recipes/install/rocky9/x86_64/xcat_stateful/slurm
 make ; %{parser} steps.tex > recipe.sh ; popd
 
@@ -175,6 +178,11 @@ install -m 0755 -p -D docs/recipes/install/%{lpath}/recipe.sh %{buildroot}/%{OHP
 install -m 0644 -p -D docs/recipes/install/%{lpath}/steps.pdf %{buildroot}/%{OHPC_PUB}/doc/recipes/%{lpath}/Install_guide.pdf
 install -m 0755 -p -D docs/recipes/install/%{lpath}/recipe.sh %{buildroot}/%{OHPC_PUB}/doc/recipes/%{lpath}/recipe.sh
 
+%define lpath rocky9/x86_64/confluent/slurm
+install -m 0644 -p -D docs/recipes/install/%{lpath}/steps.pdf %{buildroot}/%{OHPC_PUB}/doc/recipes/%{lpath}/Install_guide.pdf
+install -m 0755 -p -D docs/recipes/install/%{lpath}/recipe.sh %{buildroot}/%{OHPC_PUB}/doc/recipes/%{lpath}/recipe.sh
+
+
 %define lpath rocky9/x86_64/xcat/slurm
 install -m 0644 -p -D docs/recipes/install/%{lpath}/steps.pdf %{buildroot}/%{OHPC_PUB}/doc/recipes/%{lpath}/Install_guide.pdf
 install -m 0755 -p -D docs/recipes/install/%{lpath}/recipe.sh %{buildroot}/%{OHPC_PUB}/doc/recipes/%{lpath}/recipe.sh

docs/recipes/install/common/add_confluent_hosts_intro.tex

@@ -0,0 +1,104 @@
+%\subsubsection{Register nodes for provisioning}
+
+\noindent Next, we add {\em compute} nodes and define their properties as
+attributes in \Confluent{} database.
+These hosts are grouped logically into a group named {\em
+compute} to facilitate group-level commands used later in the recipe. The compute
+group has to be defined first before we can add any nodes to the group using the 
+{\texttt nodegroup define} command. Note the
+use of variable names for the desired compute hostnames, node IPs, MAC
+addresses, and BMC login credentials, which should be modified to accommodate
+local settings and hardware. To enable serial console access via  \Confluent{},
+{\texttt console.method}
+property is also defined. 
+
+% begin_ohpc_run
+% ohpc_validation_newline
+% ohpc_validation_comment Add hosts to cluster \ref{sec:confluent_add_nodes}
+\begin{lstlisting}[language=bash,keywords={},upquote=true,basicstyle=\footnotesize\ttfamily,]
+#define the compute group
+[sms](*\#*) nodegroupdefine compute
+
+# Define nodes as objects in confluent database
+[sms](*\#*) for ((i=0; i<$num_computes; i++)) ; do
+nodedefine ${c_name[$i]} groups=everything,compute hardwaremanagement.manager=${c_bmc[$i]} \ 
+secret.hardwaremanagementuser=${bmc_username} secret.hardwaremanagementpassword=${bmc_password} \ 
+net.hwaddr=${c_mac[$i]}
+        done
+\end{lstlisting}
+% end_ohpc_run
+
+\begin{center}
+  \begin{tcolorbox}[]
+    \small
+Defining nodes one-by-one, as done above, is only efficient
+for a small number of nodes. For larger node counts,
+\Confluent{} provides capabilities for automated detection and
+configuration.
+Consult the
+\href{https://hpc.lenovo.com/users/documentation/confluentdisco.html}{\color{blue}\Confluent{}
+Hardware Discovery \& Define Node Guide}.
+\end{tcolorbox}
+\end{center}
+
+%\clearpage
+If enabling {\em optional} IPoIB functionality (e.g. to support Lustre over \InfiniBand{}), additional
+settings are required to define the IPoIB network with \Confluent{} and specify
+desired IP settings for each compute. This can be accomplished as follows for
+the {\em ib0} interface:
+
+% begin_ohpc_run
+% ohpc_validation_newline
+% ohpc_validation_comment Setup IPoIB networking
+% ohpc_command if [[ ${enable_ipoib} -eq 1 ]];then
+% ohpc_indent 5
+\begin{lstlisting}[language=bash,keywords={},upquote=true,basicstyle=\footnotesize\ttfamily]
+# Register desired IPoIB IPs per compute
+[sms](*\#*) for ((i=0; i<$num_computes; i++)) ; do
+		nodeattrib ${c_name[i]} net.ib0.ipv4_address=${c_ipoib[i]}/${ipoib_netmask}
+        done
+\end{lstlisting}
+% ohpc_indent 0
+% ohpc_command fi
+% end_ohpc_run
+
+%\clearpage
+With the desired compute nodes and domain identified, the remaining steps in the
+provisioning configuration process are to define the provisioning mode and
+image for the {\em compute} group and use \Confluent{} commands to complete
+configuration for network services like DNS and DHCP. These tasks are
+accomplished as follows:
+
+%\clearpage
+% begin_ohpc_run
+% ohpc_validation_newline
+% ohpc_validation_comment Complete networking setup, associate provisioning image
+\begin{lstlisting}[language=bash,keywords={},upquote=true,basicstyle=\footnotesize\ttfamily,literate={BOSSHORT}{\baseosshort{}}1 {IMAGE}{\installimage{}}1]
+# Associate desired provisioning image for computes
+[sms](*\#*) nodedeploy -n compute -p rocky-9.4-x86_64-default
+\end{lstlisting}
+
+%%% If the Lustre client was enabled for computes in \S\ref{sec:lustre_client}, you
+%%% should be able to mount the file system post-boot using the fstab entry
+%%% (e.g. via ``\texttt{mount /mnt/lustre}''). Alternatively, if
+%%% you prefer to have the file system mounted automatically at boot time, a simple
+%%% postscript can be created and registered with \xCAT{} for this purpose as follows.
+%%%
+%%% % begin_ohpc_run
+%%% % ohpc_validation_newline
+%%% % ohpc_validation_comment Optionally create xCAT postscript to mount Lustre client
+%%% % ohpc_command if [ ${enable_lustre_client} -eq 1 ];then
+%%% % ohpc_indent 5
+%%% \begin{lstlisting}[language=bash,keywords={},upquote=true,basicstyle=\footnotesize\ttfamily,literate={BOSVER}{\baseos{}}1]
+%%% # Optionally create postscript to mount Lustre client at boot
+%%% [sms](*\#*) echo '#!/bin/bash' > /install/postscripts/lustre-client
+%%% [sms](*\#*) echo 'mount /mnt/lustre' >> /install/postscripts/lustre-client
+%%% [sms](*\#*) chmod 755 /install/postscripts/lustre-client
+%%% # Register script for computes
+%%% [sms](*\#*) chdef compute -p postscripts=lustre-client
+%%% \end{lstlisting}
+%%% % ohpc_indent 0
+%%% % ohpc_command fi
+%%% % end_ohpc_run
+%%%
+

docs/recipes/install/common/add_to_compute_confluent_intro.tex

@@ -0,0 +1,61 @@
+% -*- mode: latex; fill-column: 120; -*-
+
+The next step is adding \OHPC{} components to the {\em compute} nodes that at this
+point are running basic OSes.  This process will leverage two \Confluent{}-provided
+commands: \texttt{nodeshell} to run \texttt{\pkgmgr{}} installer on all the
+nodes in parallel  and \texttt{nodersync} to distribute configuration files from the
+SMS to the {\em compute} nodes.
+
+\noindent To do this, repositories on the {\em compute} nodes need to be configured
+properly.
+
+\Confluent{} has automatically setup an  OS repository on the SMS and configured the
+nodes to use it, but it has  also enabled online OS repositories.
+
+
+\noindent Next, we alse add the OHPC repo to the compute nodes \S\ref{sec:enable_repo}
+
+% begin_ohpc_run
+% ohpc_comment_header Setup nodes repositories and Install OHPC components \ref{sec:add_components}
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true]
+# Add OpenHPC repo 
+[sms](*\#*) (*\chrootinstall*) http://repos.openhpc.community/OpenHPC/3/EL_9/x86_64/ohpc-release-3-1.el9.x86_64.rpm
+\end{lstlisting}
+% end_ohpc_run
+
+The {\em compute} nodes also need access to the EPEL repository, a required
+dependency for \OHPC{} packages. 
+
+% begin_ohpc_run
+% ohpc_comment_header Configure access to EPEL repo
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true]
+# Add epel repo
+[sms](*\#*) (*\chrootinstall*) epel-release
+
+\end{lstlisting}
+% end_ohpc_run
+
+
+\noindent Additionally, a workaround is needed for \OHPC{} documentation files,
+which are installed into a read-only NFS share /opt/ohpc/pub. Any package
+attempting to write to that directory will fail to install. The following
+prevents that by directing \texttt{rpm} not to install documentation files on
+the {\em compute} nodes:
+
+% begin_ohpc_run
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true]
+[sms](*\#*) nodeshell compute echo -e %_excludedocs 1 \>\> ~/.rpmmacros
+\end{lstlisting}
+% end_ohpc_run
+
+\noindent Now \OHPC{} and other cluster-related software components can be
+installed on the nodes. The first step is to install a base compute package:
+% begin_ohpc_run
+% ohpc_comment_header Add OpenHPC base components to compute image
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true]
+# Install compute node base meta-package
+[sms](*\#*) (*\chrootinstall*) ohpc-base-compute
+\end{lstlisting}
+% end_ohpc_run
+
+\noindent Next, we can include additional components:

docs/recipes/install/common/confluent_compute_bos_intro.tex

@@ -0,0 +1,4 @@
+Installation is accomplished in two steps: First,  a generic OS
+image is installed on {\em compute} nodes and then, once the nodes are up
+and running, \OHPC{} components are added to both the SMS and the nodes at the
+same time.

docs/recipes/install/common/confluent_customize.tex

@@ -0,0 +1,35 @@
+Here we set up \NFS{}  mounting of a
+\$HOME file system and the public \OHPC{} install path (\texttt{/opt/ohpc/pub})
+that will be hosted by the {\em master} host in this  example configuration.
+
+\vspace*{0.15cm}
+% begin_ohpc_run
+% ohpc_comment_header Customize system configuration \ref{sec:master_customization}
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true]
+# Disable /tftpboot and /install export entries
+[sms](*\#*) perl -pi -e "s|/tftpboot|#/tftpboot|" /etc/exports
+[sms](*\#*) perl -pi -e "s|/install|#/install|" /etc/exports
+
+# Export /home and OpenHPC public packages from master server
+[sms](*\#*) echo "/home *(rw,no_subtree_check,fsid=10,no_root_squash)" >> /etc/exports
+[sms](*\#*) echo "/opt/ohpc/pub *(ro,no_subtree_check,fsid=11)" >> /etc/exports
+[sms](*\#*) exportfs -a
+[sms](*\#*) systemctl restart nfs-server
+[sms](*\#*) systemctl enable nfs-server
+
+
+# Create NFS client mounts of /home and /opt/ohpc/pub on compute hosts
+[sms](*\#*) nodeshell compute echo \
+        "\""${sms_ip}:/home /home nfs nfsvers=3,nodev,nosuid 0 0"\"" \>\> /etc/fstab
+[sms](*\#*) nodeshell compute echo \
+        "\""${sms_ip}:/opt/ohpc/pub /opt/ohpc/pub nfs nfsvers=3,nodev 0 0"\"" \>\> /etc/fstab
+[sms](*\#*) nodeshell compute systemctl restart nfs
+
+# Mount NFS shares
+[sms](*\#*) nodeshell compute mount /home
+[sms](*\#*) nodeshell compute mkdir -p /opt/ohpc/pub
+[sms](*\#*) nodeshell compute mount /opt/ohpc/pub
+
+\end{lstlisting}
+% end_ohpc_run
+

docs/recipes/install/common/confluent_init_os_images_rocky.tex

@@ -0,0 +1,59 @@
+% -*- mode: latex; fill-column: 120; -*-
+
+With the provisioning services enabled, the next step is to define
+a system image that can subsequently be
+used to provision one or more {\em compute} nodes. The following subsections highlight this process.
+
+\subsubsection{Build initial BOS image} \label{sec:assemble_bos}
+The following steps illustrate the process to build a minimal, default image for use with \Confluent{}. To begin, you will
+first need to have a local copy of the ISO image available for the underlying OS. In this recipe, the relevant ISO image
+is \texttt{Rocky-9.4-x86\_64-dvd.iso} (available from the Rocky
+\href{https://rockylinux.org/download/}{\color{blue}download} page).
+We initialize the image
+creation process using the \texttt{osdeploy} command assuming that the necessary ISO image is available locally in
+\texttt{\$\{iso\_path\}} as follows:
+
+The \texttt{osdeploy initialize} command is used to prepare a confluent server to deploy deploy operating systems.
+For first time setup, run osdeploy initialize interactively to be walked through the various options using: 
+\texttt{osdeploy initialize -i}
+
+% begin_ohpc_run
+% ohpc_comment_header Initialize OS images for use with Confluent \ref{sec:assemble_bos}
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true,keepspaces,literate={BOSVER}{\baseos{}}1]
+[sms](*\#*) osdeploy initialize -${initialize_options}
+[sms](*\#*) osdeploy import ${iso_path}
+
+\end{lstlisting}
+% end_ohpc_run
+
+\noindent Once completed, OS image should be available for use within \Confluent{}. These can be queried via:
+
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true,keepspaces,literate={BOSVER}{\baseos{}}1]
+# Query available images
+[sms](*\#*) osdeploy list
+Distributions:
+  rocky-8.5-x86_64
+  rocky-9.4-x86_64
+Profiles:
+  rhel-9.4-x86_64-default
+  rocky-8.5-x86_64-default
+\end{lstlisting}
+
+If needing to copy files from the sms node to the compute nodes during deployment, this can be done by
+modifying the syncfiles file that is created when \texttt{osdeploy import} command is run. For an environment
+that has no DNS server and needs to have /etc/hosts file synced amongst all the nodes, the following command
+should be run.   
+
+% begin_ohpc_run
+% ohpc_validation_newline
+% ohpc_validation_comment Sync the hosts file in cluster
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true,keepspaces,literate={BOSVER}{\baseos{}}1]
+  [sms](*\#*) echo "/etc/hosts -> /etc/hosts" >> /var/lib/confluent/public/os/rocky-9.4-x86_64-default/syncfiles
+
+\end{lstlisting}
+% end_ohpc_run
+
+
+%The \texttt{CHROOT} environment variable highlights the path and is used by
+%subsequent commands to augment the basic installation.
+

docs/recipes/install/common/confluent_setup.tex

@@ -0,0 +1,32 @@
+At this point, all of the packages necessary to use \Confluent{} on the {\em master}
+host should be installed. Next, we enable support for local provisioning using
+a second private interface (refer to Figure~\ref{fig:physical_arch})
+
+% begin_ohpc_run
+% ohpc_comment_header Complete basic Confluent setup for master node \ref{sec:setup_confluent}
+%\begin{verbatim}
+\begin{lstlisting}[language=bash,literate={-}{-}1,keywords={},upquote=true,keepspaces]
+# Enable internal interface for provisioning
+[sms](*\#*) ip link set dev ${sms_eth_internal} up
+[sms](*\#*) ip address add ${sms_ip}/${internal_netmask} broadcast + dev ${sms_eth_internal}
+
+\end{lstlisting}
+%\end{verbatim}
+% end_ohpc_run
+
+
+\noindent \Confluent{} requires a network domain name specification for system-wide name
+resolution. This value can be set to match your local DNS schema or given a
+unique identifier such as `local`. A default group called everything is 
+automatically added to every node. It provides a method to indicate global settings.
+Attributes may all be specified on the command line, and an example set could be:
+
+% begin_ohpc_run
+% ohpc_validation_newline
+% ohpc_validation_comment Define local domainname, deployment protocol and dns 
+\begin{lstlisting}[language=bash,keywords={},upquote=true,basicstyle=\footnotesize\ttfamily,literate={BOSVER}{\baseos{}}1]
+[sms](*\#*) nodegroupattrib everything deployment.useinsecureprotocols=${deployment_protocols} dns.domain=${dns_domain} 
+[sms](*\#*) nodegroupattrib everything dns.servers=${dns_servers} net.ipv4_gateway=${sms_ip}
+\end{lstlisting}
+
+\noindent We will also define