The master daemon, bpmaster is the central part of BProc system. It runs on the front-end machine. Once it is running, the slave nodes run the slave daemon, bpslave to connect to the front-end machine. bpmaster runs on the front-end machine and handles all the details of running BProc.

> bpmaster

down Nodes are `down' when they are NOT connected to the BProc master daemon. It is impossible to do anything to nodes via BProc when they are in this state.

unavailable When a node is `unavailable', it is connected to the BProc master but has not yet been tagged as ready for users. While in this state, the system makes no guarantees about the state of the node. Unavailable usually means that the node is in some transitional state. The node may be booting and setting up or it may be shutting down. It is also possible that the system administrator has manually set the node state to unavailable to indicate that it should not be used for some other reason. Nodes also remain unavailable after booting if their hard disks have not been partitioned.

up When a node is tagged as `up', it is available for use by users.

error A node that does not boot successfully is marked with `error'. `error' may also indicate an unpartitioned slave node hard disk.

reboot It is possible to `reboot' a node.

halt It is possible to `halt' a node (suspend processing, but not power off).

pwroff It is possible to remotely power off a node using `pwroff'.

Node states may be viewed and manually manipulated using the bpctl program.

VMADump is the system used by BProc to take a running process and copy it to a remote node. VMADump saves or restores a process's memory space to or from a stream. In the case of BProc, the stream is a TCP socket to the remote machine. VMADump implements an optimization which greatly reduces the size of the memory space. Most programs on the system are dynamically linked. At run time, they will use mmap to get copies of various libraries in their memory spaces. Since they are demand paged, the entire library is always mapped even if most of it will never be used. These regions must be included when copying a process's memory space and again when the process is restored. This is expensive since the C library dwarfs most programs in size. Here is an example memory space for the program sleep. This is taken directly from `/proc/pid/maps'.

08048000-08049000 r-xp 00000000 03:01 288816     /bin/sleep
08049000-0804a000 rw-p 00000000 03:01 288816     /bin/sleep
40000000-40012000 r-xp 00000000 03:01 911381     /lib/ld-2.1.2.so
40012000-40013000 rw-p 00012000 03:01 911381     /lib/ld-2.1.2.so
40017000-40102000 r-xp 00000000 03:01 911434     /lib/libc-2.1.2.so
40102000-40106000 rw-p 000ea000 03:01 911434     /lib/libc-2.1.2.so
40106000-4010a000 rw-p 00000000 00:00 0
bfffe000-c0000000 rwxp fffff000 00:00 0

The total size of the memory space for this trivial program is 1089536 bytes. All but 32K of that comes from shared libraries - VMADump takes advantage this. Instead of storing the data contained in each of these regions, it stores a reference to the regions. When the image is restored, that files will be mmaped to the same memory locations. In order for this optimization to work, VMADump must know which files it can expect to find in the location where they are restored. VMADump has a list of files which it presumes are present on remote systems. The vmadlib utility exists to manage this list. Note that VMADump will correctly handle regions mapped with MAP_PRIVATE, which have been written. VMADump does not specially handle shared memory regions. A copy of the data within the region will be included in the dump. No attempt to re-share the region will be made at restoration time. The process will get a private copy. VMADump does not save or restore any information about file descriptors. VMADump will only dump a single thread of a multi-threaded program. There is currently no way to dump a multi-threaded program in a single dump.

Basic usage information for the binaries that are included with BProc.

bpmaster

bpmaster -h
bpmaster -v
bpmaster [ -d ] [ -c c_file ] [ -m m_file ]

bpmaster is the BProc master daemon. It runs on the front-end machine of a cluster running BProc. It listens on a TCP port and accepts connections from slave daemons. Configuration information comes from the Beowulf configuration file. The BProc master daemon reads interface, iprange, bprocport, allowinsecureports and logfacility.

Options:

@option{-h}

Display a help message and exit.

@option{-v}

Display version information and exit.

@option{-d}

Increase debugging (verbose)

@option{-c c_file}

Read configuration information from c_file instead of the default file (`/etc/beowulf/config').

@option{-m m_file}

Dump a message trace to m_file. This is only useful for debugging but slows down the daemons.

bpslave

bpslave -h
bpslave [ -l log ] [ -r ] [ -d ] [ -m m_file ] [ -v ] masterhostname port

bpslave is the BProc slave daemon. It runs on slave nodes in a cluster and connects to the front-end machine (masterhostname) to accept jobs through masterport (port).

Options:

@option{-h}

Display a help message and exit.

@option{-l log}

Specify the log facility to which the messages should be sent (default=daemon).

@option{-r}

Automatically reconnect if the connection to the master daemon is lost.

@option{-d}

Do not daemonize self.

@option{-m m_file}

Dump a message trace to m_file. This is only useful for debugging but slows down the daemons.

@option{-v}

Increase verbose level (implies -d)

bpstat

bpstat -h
bpstat -v
bpstat -n
bpstat -u
bpstat [ -a nodenum ]
bpstat [ -s nodenum ]
bpstat -m
bpstat -p
bpstat -P

bpstat displays various pieces of status information about cluster nodes. The display is formatted in columns specifying node number, node address, node status, user access and group access. This program also includes a number of options intended to be useful for scripts.

Options:

@option{-h}

Display a help message and exit.

@option{-v}

Display version information and exit.

@option{-n}

Print the number of nodes in the machine. Note that this is the number of nodes configured (via iprange) not the number of nodes that are up.

@option{-u}

Print the number of nodes that are up.

@option{-a nodenum}

Print the IP address of nodenum, where nodenum is value 0 through N-1 or -1 (the front-end machine) or -2 (the current node).

@option{-s nodenum}

Print the status of nodenum, where nodenum is value 0 through N-1 or -1 (the front-end machine) or -2 (the current node).

@option{-m}

Display machine state. (This is the default mode of operation.)

@option{-p}

Display process state.

@option{-P}

Read output from @command{ps} from standard input and add a column containing the node which the process exists on.

bpctl

bpctl -h
bpctl -v
bpctl -M [ -a ] commands
bpctl -S nodenum [ -a ] [ -r dir ] [ -s state ] commands

bpctl is bproc control. Used to apply commands to referenced nodes.

Options:

@option{-h}

Display a help message and exit.

@option{-v}

Display version information and exit.

@option{-M commands}

Apply the commands to the front-end machine.

@option{-S nodenum commands}

where nodenum is value, 0 through N-1 or all. Apply the following commands to slave node, nodenum.

@option{-a}

Print the IP address of the front-end machine (when used with the -M option). Print the IP address of the slave node (when used with the -S option).

@option{-r dir}

Ask the slave daemon to perform a chroot() to dir. After doing this, all processes started on a node via BProc will see dir as their root directory. This command is only usable on slave nodes.

@option{-s state}

Set slave state to state. The valid node states are `down', `unavailable', `error', `up', `reboot', `halt' and `pwroff'. The state of nodes in the `down' state cannot be changed. Setting the state of a node to `down' will cause a node to be disconnected from the master daemon.

bpsh

bpsh -h
bpsh -v
bpsh [-n] nodenum command
bpsh -a [-n] command
bpsh -A [-n] command

bpsh is a rsh replacement. It runs command on nodenum. nodenum may be a single node number, a comma delimited list of nodes, -a for all nodes which are up or -A for all nodes which are not down. (Nodes which are not down may be in the error or unavailable states.)

bpsh will forward standard input, standard output and standard error for the remote processes. Standard output and error are forwarded subject to the flags below. Standard input will be forwarded to the remote processes. If there is more than one remote process, standard input will be duplicated for every remote node.

For a single remote process, the exit status of bpsh will be the exit status of the remote process. Note that this can include non-normal exit status such as being killed by a signal.

When there are many remote processes, bpsh exits with the highest exit status of all the remote processes.

Options:

@option{-h}

Display a help message and exit.

@option{-v}

Display version information and exit.

@option{-a}

Run the command on all nodes which are `up'.

@option{-A}

Run the command on all possible nodes (those marked `up', `unavailable' or `error')

@option{-n}

Redirect stdin from /dev/null.

@option{-L}

Line buffer output from remote nodes.

@option{-p}

Prefix each line of output with the node number it is from. This implies @option{-L}.

@option{-s}

Show the output from each node sequentially. By default the output from different nodes can be mixed in any order.

@option{-d}

Print a divider between the output from each node. This implies @option{-s}.

@option{-b bytes}

Set the IO buffer size to bytes. This affects the maximum line length for line buffered IO. The default value is 4096.

bpcp

bpcp -h
bpcp -v
bpcp [ -p ] f1 f2
bpcp [ -r ] [ -p ] f1 ... fn dir

bpcp copies files between machines. Each file (f1...fn)or directory argument (dir) is either a remote file name of the form node:path, or a local file name (containing no `:' characters).

Options:

@option{-h}

Display a help message and exit.

@option{-v}

Display version information and exit.

@option{-p}

Preserve file timestamps.

@option{-r}

Copy directories recursively.

vmalib

vmadlib -c
vmadlib -a [ libs ... ]
vmadlib -d [ libs ... ]
vmadlib -l

This program is a utility to manage the VMADump in-kernel library list.

Options:

@option{-c}

Clear the library list.

@option{-a [ libs ... ]}

Add libs to the library list. If `-' is given as an argument, newline separated library file names will read from standard input.

@option{-d [ libs ... ]}

Delete libs to the library list. If `-' is given as an argument, newline separated library file names will read from standard input.

@option{-l}

List the libraries in the library list. 

bWatch:

After BPROC is installed the installation of bWatch a monitoring utility can be done. Download bWatch-1.0.2.tar. Issue the comand gunzip -c bWatch-1.0.2.tar | tar xovf -. Then in the bWatch-1.0.2 directory check that the path to "wish" is correct. This is usually /usr/bin/ or /usr/local/bin/. This is important since bWatch is a Tcl/Tk script and will not install or run withour "wish". Then type make and make install. After this in the $HOME directory of the root and usr edit the .bWatchrc.tcl so that the entry set listOfhost is as shown below:

set listOfHost {master

                       node2

                       node3

                       node4}

Note if you are using rsh then the .rhosts file in the root and user must use the same names.

Then if you run bWatch.tcl you will see the windows as shown below under GNOME.

This completes the installation of BPROC with bWatch.

BACK TO TOP

BACK TO MICROWAY APPLICATION NOTE INDEX