LDP
/
LDP
mirror of https://github.com/tLDP/LDP
0
1
Fork 0
Code Releases Activity
LDP/LDP/howto/docbook/openMosix-HOWTO/openMosix_Admin.sgml

466 lines
14 KiB
Plaintext
Raw Blame History

<CHAPTER ID="Admin">
<TITLE>Administrating openMosix</TITLE>
<SECT1><TITLE>Basic Administration</TITLE>
<PARA>openMosix provides the advantage of process migration to HPC-applications.
The administrator can configure and tune the openMosix-cluster by using
the openMosix-user-space-tools
or the /proc/hpc interface which will be now described in detail.
</PARA>
<para>
Up till openMosix version 2.4.16 the /proc interface was named
/proc/mosix !
Until openMosix version 2.4.17 it was named /proc/hpc.
</para>
</SECT1><SECT1><TITLE>Configuration</TITLE>
<PARA>
The values in the flat files in the /proc/hpc/admin directory presenting
the current configuration of
the cluster. Also the administrator can write its own values into these
files to change the configuration
during runtime, e.g.
</para>
<table frame=all><title>Changing /proc/hpc parameters</title>
<tgroup cols=2 align=left>
<tbody>
<row><entry>
echo 1 > /proc/hpc/admin/block</entry><entry>blocks the arrival of remote processes</entry>
</row>
<row>
<entry>
echo 1 > /proc/hpc/admin/bring</entry><entry>bring all migrated processes home</entry>
</row>
</tbody></tgroup>
</table>
<para>
...
</para>
<table frame=all><title>/proc/hpc/admin/</title>
<tgroup cols=3 align=left>
<tbody>
<row><entry>
(binary files) </entry><entry> config </entry><entry>
the main configuration file (written by the setpe util)</entry></row>
<row><entry>
(flat files)</entry><entry> block </entry><entry> allow/forbid arrival of remote processes</entry></row>
<row><entry></entry><entry> bring </entry><entry> bring home all migrated processes
</entry></row>
<row><entry></entry><entry>
dfsalinks </entry><entry> list of current symbolic dfsa-links
</entry></row>
<row><entry></entry><entry>
expel </entry><entry> sending guest processes home
</entry></row>
<row><entry></entry><entry>
gateways </entry><entry> maximum number of gateways
</entry></row>
<row><entry></entry><entry>
lstay </entry><entry> local processes should stay
</entry></row>
<row><entry></entry><entry>
mospe </entry><entry> contains the openMosix node id
</entry></row>
<row><entry></entry><entry>
nomfs </entry><entry> disables/enables MFS
</entry></row>
<row><entry></entry><entry>
overheads </entry><entry> for tuning
</entry></row>
<row><entry></entry><entry>
quiet </entry><entry> stop collecting load-load-balancing informations
</entry></row>
<row><entry></entry><entry>
decay-interval </entry><entry> interval for collecting informations about load-balancing
</entry></row>
<row><entry></entry><entry>
slow-decay </entry><entry> default 975
</entry></row>
<row><entry></entry><entry>
fast-decay </entry><entry> default 926
</entry></row>
<row><entry></entry><entry>
speed </entry><entry> speed relative to PIII/1GHz)
</entry></row>
<row><entry></entry><entry>
stay </entry><entry> enables/disables automatic process migration
</entry></row>
</tbody></tgroup></table>
<table frame=all><title>Writing a 1 to the following files /proc/hpc/decay/</title>
<tgroup cols=2 align=left>
<tbody>
<row><entry>
clear </entry><entry> clears the decay statistics
</entry></row>
<row><entry>
cpujob </entry><entry> tells openMosix that the process is cpu-bound
</entry></row>
<row><entry>
iojob </entry><entry> tells openMosix that the process is io-bound
</entry></row>
<row><entry>
slow </entry><entry> tells openMosix to decay its statistics slow
</entry></row>
<row><entry>
fast </entry><entry> tells openMosix to decay its statistics fast
</entry></row>
</tbody></tgroup></table>
<table frame=all><TITLE>Informations about the other nodes</TITLE>
<tgroup cols=2 align=left>
<tbody>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/CPUs </entry><entry> how many CPU's the node has
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/load </entry><entry> the openMosix load of this node
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/mem </entry><entry> available memory as openMosix believes
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/rmem </entry><entry> available memory as Linux believes
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/speed </entry><entry> speed of the node relative to PIII/1GHz
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/status </entry><entry> status of the node
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/tmem </entry><entry> available memory
</entry></row>
<row><entry>
/proc/hpc/nodes/[openMosix_ID]/util </entry><entry> utilization of the node
</entry></row>
</tbody>
</tgroup></table>
<table frame=all><title>
Additional Informations about local processes
</title>
<tgroup cols=2 align=left>
<tbody>
<row><entry>
/proc/[PID]/cantmove </entry><entry> reason why a process cannot be migrated
</entry></row>
<row><entry>
/proc/[PID]/goto </entry><entry> to which node the process should migrate
</entry></row>
<row><entry>
/proc/[PID]/lock </entry><entry>if a process is locked to its home node
</entry></row>
<row><entry>
/proc/[PID]/nmigs </entry><entry>how many times the process migrated
</entry></row>
<row><entry>
/proc/[PID]/where </entry><entry>where the process is currently being computed
</entry></row>
<row><entry>
/proc/[PID]/migrate </entry><entry>same as goto remote processes
</entry></row>
<row><entry>
/proc/hpc/remote/from </entry><entry>the home node of the process
</entry></row>
<row><entry>
/proc/hpc/remote/identity </entry><entry>additional informations about the process
</entry></row>
<row><entry>
/proc/hpc/remote/statm </entry><entry>memory statistic of the process
</entry></row>
<row><entry>
/proc/hpc/remote/stats </entry><entry>cpu statistics of the process
</entry></row>
</tbody></tgroup></table>
</SECT1>
<SECT1><TITLE>the userspace-tools</TITLE>
<PARA>
These following tools are providing easy administration to openMosix
clusters.
<PROGRAMLISTING>
migrate -send a migrate request to a process
syntax:
migrate [PID] [openMosix_ID]
</PROGRAMLISTING>
<PROGRAMLISTING>
mon -is a ncurses-based terminal monitor
several informations about the current status are displayed in bar-charts
</PROGRAMLISTING>
<PROGRAMLISTING>
mosctl -is the openMosix main configuration utility
syntax:
mosctl [stay|nostay]
[lstay|nolstay]
[block|noblock]
[quiet|noquiet]
[nomfs|mfs]
[expel|bring]
[gettune|getyard|getdecay]
mosctl whois [openMosix_ID|IP-address|hostname]
mosctl [getload|getspeed|status|isup|getmem|getfree|getutil] [openMosix_ID]
mosctl setyard [Processor-Type|openMosix_ID||this]
mosctl setspeed interger-value
mosctl setdecay interval [slow fast]
</PROGRAMLISTING>
<table frame=all><title>more detailed</title>
<tgroup cols=2 align=left>
<tbody><row><entry>
stay </entry><entry>no automatic process migration
</entry></row>
<row><entry>
nostay </entry><entry>automatic process migration (default)
</entry></row>
<row><entry>
lstay </entry><entry>local processes should stay
</entry></row>
<row><entry>
nolstay </entry><entry>local processes could migrate
</entry></row>
<row><entry>
block </entry><entry>block arriving of guest processes
</entry></row>
<row><entry>
noblock </entry><entry>allow arriving of guest processes
</entry></row>
<row><entry>
quiet </entry><entry>disable gathering of load-balancing informations
</entry></row>
<row><entry>
noquiet </entry><entry>enable gathering of load-balancing informations
</entry></row>
<row><entry>
nomfs </entry><entry>disables MFS
</entry></row>
<row><entry>
mfs </entry><entry>enables MFS
</entry></row>
<row><entry>
expel </entry><entry>send away guest processes
</entry></row>
<row><entry>
bring </entry><entry>bring all migrated processes home
</entry></row>
<row><entry>
gettune </entry><entry>shows the current overhead parameter
</entry></row>
<row><entry>
getyard </entry><entry>shows the current used Yardstick
</entry></row>
<row><entry>
getdecay </entry><entry>shows the current decay parameter
</entry></row>
<row><entry>
whois </entry><entry>resolves openMosix-ID, ip-addresses and hostnames of the cluster
</entry></row>
<row><entry>
getload </entry><entry>display the (openMosix-) load
</entry></row>
<row><entry>
getspeed </entry><entry>shows the (openMosix-) speed
</entry></row>
<row><entry>
status </entry><entry>displays the current status and configuration
</entry></row>
<row><entry>
isup </entry><entry>is a node up or down (openMosix kind of ping)
</entry></row>
<row><entry>
getmem </entry><entry>shows logical free memory
</entry></row>
<row><entry>
getfree </entry><entry>shows physical free mem
</entry></row>
<row><entry>
getutil </entry><entry>display utilization
</entry></row>
<row><entry>
setyard </entry><entry>sets a new Yardstick-value
</entry></row>
<row><entry>
setspeed </entry><entry>sets a new (openMosix-) speed value
</entry></row>
<row><entry>
setdecay </entry><entry>sets a new decay-interval
</entry></row>
</tbody></tgroup></table>
<PROGRAMLISTING>
mosrun -run a special configured command on a chosen node
syntax:
mosrun [-h|openMosix_ID| list_of_openMosix_IDs] command [arguments]
</PROGRAMLISTING>
</PARA><PARA>
The mosrun command can be executed with several more commandline options.
To ease this up there are several preconfigured run-scripts for executing
jobs with a special (openMosix) configuration.
</PARA>
<table frame=all><title>extra options for mosrun</title>
<tgroup cols=2 align=left>
<tbody><row><entry>
nomig </entry><entry>runs a command which process(es) won't migrate
</entry></row>
<row><entry>
runhome </entry><entry>executes a command locked to its home node
</entry></row>
<row><entry>
runon </entry><entry>runs a command which will be directly migrated and locked to a node
</entry></row>
<row><entry>
cpujob </entry><entry>tells the openMosix cluster that this is a cpu-bound process
</entry></row>
<row><entry>
iojob </entry><entry>tells the openMosix cluster that this is a io-bound process
</entry></row>
<row><entry>
nodecay </entry><entry>executes a command and tells the cluster not to refresh the load-balancing statistics
</entry></row>
<row><entry>
slowdecay </entry><entry>executes a command with a slow decay interval for collecting load-balancing statistics
</entry></row>
<row><entry>
fastdecay </entry><entry>executes a command with a fast decay interval for collecting load-balancing statistics
</entry></row>
</tbody></tgroup></table>
<PROGRAMLISTING>
setpe -manual node configuration utility
syntax:
setpe -w -f [hpc_map]
setpe -r [-f [hpc_map]]
setpe -off
-w reads the openMosix configuration from a file (typically /etc/hpc.map)
-r writes the current openMosix configuration to a file (typically /etc/hpc.map)
-off turns the current openMosix configuration off
</PROGRAMLISTING>
<PROGRAMLISTING>
tune openMosix calibration and optimizations utility.
(for further informations review the tune-man page)
</PROGRAMLISTING>
<PARA>
Additional to the /proc interface and the commandline-openMosix utilities
(which are using the
/proc interface) there is a patched "ps" and "top" available (they are
called "mps" and "mtop")
which displays also the openMosix-node ID on a column. This is useful for
finding out where a
specific process is currently being computed.
</PARA>
<!--
<PARA>
The administrator can have a overview about the current status of the
cluster and its nodes with
the "Mosix Cluster Information Tool PHP" which can be found at
<ulink url="http://wijnkist.warande.uu.nl/mosix/"><citetitle> http://wijnkist.warande.uu.nl/mosix/
</citetitle></ulink>
.
(the path to the NODESDIR has to be adjusted to
$NODESDIR="/proc/hpc/nodes/")
</PARA> -->
<PARA>
This actually summarised the command line tools, but have a look at
openMosixview which is
a
GUI for the most common
administration tasks, and which ill be discussed in a future chapter.
</PARA>
</SECT1>
<sect1><title>Cluster Mask</title>
<para>
(by Moshe Bar)
</para><para>
Several people have asked for a feature in openMosix which allows to
specifiy to which nodes a given process and it's children can migrate
and to which nodes it cannot.
</para>
<para>
Simone Ettore has just committed a new patch to the CVS which allows
you to do just that.
</para>
<para>
Here is how it works:
<itemizedlist><listitem>
<para>
/proc/[pid]/migfilter enable/disable the capability of filter migration.
</para>
</listitem>
<listitem>
<para>
/proc/[pid]/mignodes is a bit-list of nodes. The bit position of a node
is calculated as 2^(PE-1). PE is node number.
</para></listitem>
<listitem><para>
/proc/[pid]/migpolicy is the policy of the filtering:
0=DENY: the process can migrate in all nodes except when the relative
bit on mignodes is 1
1=ALLOW: the process can migrate in all nodes where the relative bit on
mignodes is 1
</para>
</listitem>
</itemizedlist>
</para>
<para>
We are shortly going to release also a simple user-land tool to set the
node mask, but I would like you guys to give it a try asap before we
release it as openMosix 2.4.20-3.
</para>
</sect1>
</CHAPTER>
View Git Blame Copy Permalink