4. How to on solaris resource management?

How to on solaris resource management?


  • Here is a basic write up on how to setup resource management for apache webserver.

    Check project is set to correct lookup entry

    # grep project /etc/nsswitch.conf
project: files

    Add a project for daemon group

    # projadd -c "Apache default" group.daemon
# cat /etc/project
system:0::::
user.root:1::::
noproject:2::::
default:3::::
group.staff:10::::
group.daemon:100:Apache default:::

    Check the project id has been set

    # su - daemon
$ pwd
/
$ id
uid=1(daemon) gid=1(other)
$ id -p
uid=1(daemon) gid=1(other) projid=3(default)

    Change parameters for system V IPC:

    # projmod -sK "project.max-shm-memory=(privileged,200m,
deny)" group.daemon
# cat /etc/project
system:0::::
user.root:1::::
noproject:2::::
default:3::::
group.staff:10::::
group.daemon:100:Apache default:::project.max-shm-memory=(privileged,209715200,deny)

    Verify the IPC values has been set

    # su - daemon
$
$ id -p
uid=1(daemon) gid=1(other) projid=3(default)

$ prctl -n project.max-shm-memory -i process $$
process: 27265: -su
NAME PRIVILEGE VALUE FLAG ACTION RECIPIENT
project.max-shm-memory
privileged 126MB - deny -

    The startup of groups of processes in projects must be project aware on startup.This is done via SMF.

    For example two DB instances salesdb and mktdb should have separate service during startup.

    The newtask -p command must tell the instance to use the assigned project.

    Check the status of apache

    # svcs -a | grep apache2
disabled 18:55:33 svc:/network/http:apache2

    Enabling the service

    # svcadm enable -r svc:/network/http:apache2

    Check status again, still in maintenance mode, there's a problem which needs looking at.

    # svcs -a | grep apache2
maintenance 18:07:45 svc:/network/http:apache2

    Check the service logs

    # svcs -xv apache2
svc:/network/http:apache2 (Apache 2 HTTP server)
State: maintenance since Sat Aug 16 20:48:13 2008
Reason: Start method exited with $SMF_EXIT_ERR_CONFIG.
See: http://sun.com/msg/SMF-8000-KS
See: man -M /usr/apache2/man -s 8 httpd
See: /var/svc/log/network-http:apache2.log
Impact: This service is not running.

    In this case no httpd.conf file existed for copied from standard template.

    # cd /etc/apache2
# ls
highperformance-std.conf httpd.conf-example ssl-std.conf
highperformance.conf magic ssl.conf
httpd.conf mime.types
# mv httpd-std.conf httpd.conf

    Edit the httpd.conf file accordingly, solaris does not
    like group ids over certain numbers, so change the group
    to something else, I've changed it to have group webservd
    and user to webservd.

    Status will still be in maintenance state.

    # svcs -a | grep apache
legacy_run 18:56:10 lrc:/etc/rc3_d/S50apache
maintenance 20:48:13 svc:/network/http:apache2

    First the service needs to be cleared and then check to see if the maintenance state has gone.

    # svcadm clear svc:/network/http:apache2

    In this case it gone, the next state on enable should be online, currently in an offline state.

    # svcs -l svc:/network/http:apache2
fmri svc:/network/http:apache2
name Apache 2 HTTP server
enabled true
state offline
next_state online
state_time Sat Aug 16 21:00:00 2008
logfile /var/svc/log/network-http:apache2.log
restarter svc:/system/svc/restarter:default
contract_id 112
dependency require_all/error svc:/milestone/network:default (online)
dependency require_all/none svc:/system/filesystem/local:default (online)
dependency optional_all/error svc:/system/filesystem/autofs:default (online)

    Check status.

    # svcs apache2
STATE STIME FMRI
offline* 21:00:46 svc:/network/http:apache2

    Enable the service and check status again.

    # svcadm enable svc:/network/http:apache2
# svcs apache2
STATE STIME FMRI
online 21:13:09 svc:/network/http:apache2

    Do a ps check to make sure:

    # ps -ef | grep httpd
webservd 9026 9023 0 21:17:10 ? 0:00 /usr/apache2/bin/httpd -k start
webservd 9027 9023 0 21:17:10 ? 0:00 /usr/apache2/bin/httpd -k start
root 9023 1 0 21:17:09 ? 0:00 /usr/apache2/bin/httpd -k start
webservd 9024 9023 0 21:17:10 ? 0:00 /usr/apache2/bin/httpd -k start
webservd 9025 9023 0 21:17:10 ? 0:00 /usr/apache2/bin/httpd -k start
webservd 9028 9023 0 21:17:10 ? 0:00 /usr/apache2/bin/httpd -k start

    Add an apache project.

    # projadd -c "Apache Default" group.webservd
# cat /etc/project
system:0::::
user.root:1::::
noproject:2::::
default:3::::
group.staff:10::::
group.webservd:101:Apache Default:::

    Define the amount of shared memory

    # projmod -sK "project.max-shm-memory=(privileged,200m,deny)" group.webservd
# cat /etc/project
system:0::::
user.root:1::::
noproject:2::::
default:3::::
group.staff:10::::
group.webservd:101:Apache Default:::project.max-shm-memory=(privileged,209715200,deny)

    Edit the manifest file for apach2.

    # cd /var/svc/manifest/network
# vi http-apache2.xml

    Add the following lines in the start , note the timeout
    line is already there but remove the slash, and change
    the names to the user and project.

    Run the xmlchecker to verify syntax, prints the entire file if ok, otherwise there'll be errors to fix.

    # xmllint http-apache2.xml

    Import the changes to the manifest.

    # svccfg import http-apache2.xml

    Change the owner and group of the logs files.

    # cd /var/apache2/logs
# ls -l
total 928
-rw-r--r-- 1 root root 0 Aug 16 20:59 access_log
-rw-r--r-- 1 webservd root 465064 Aug 17 10:06 error_log
# chown webservd:webservd access_log error_log

    Change the owner and group of the apache run directory

    # ls -ld /var/run/apache2/
drwxr-xr-x 2 root root 69 Aug 17 10:13 /var/run/apache2/
# chown webservd:webservd /var/run/apache2/
# cd /var/run/apache2/
# ls -l
total 0

    Disable and enable the apache2 service.

    # svcadm disable apache2
# svcadm enable apache2

    Verify project is set on user webservd

    # su - webservd
$ id -p
uid=80(webservd) gid=80(webservd) projid=102(group.webservd)

    Verify webserver processes are running under the new project

    # ps -fu webservd -o user,project,comm
USER PROJECT COMMAND
webservd group.webservd /usr/apache2/bin/httpd
webservd group.webservd /usr/apache2/bin/httpd
webservd group.webservd /usr/apache2/bin/httpd
webservd group.webservd /usr/apache2/bin/httpd
webservd group.webservd /usr/apache2/bin/httpd
webservd group.webservd /usr/apache2/bin/httpd

    Assign 20 shares of the cpu to group.

    # prctl -n project.cpu-shares -r -v 20 -i project group.webservd

    Verify & display the cpu shares setting.

    # prctl -n project.cpu-shares -i project group.webservd
project: 102: group.webservd
NAME PRIVILEGE VALUE FLAG ACTION RECIPIENT
project.cpu-shares
privileged 20 - none -
system 65.5K max none -

    Make the changes to the /etc/project file.

    # projmod -sK "project.cpu-shares=(privileged,20,none)" group.webservd
# cat /etc/project
system:0::::
user.root:1::::
noproject:2::::
default:3::::
group.staff:10::::
group.webservd:102:Apache Default:::project.cpu-shares=(privileged,20,none);project.
max-shm-memory=(privileged,209715200,deny)

    Activate resource pools

    # pooladm -e

    List resource pool information.

    # pooladm

system default
string system.comment
int system.version 1
boolean system.bind-default true
string system.poold.objectives wt-load

pool pool_default
int pool.sys_id 0
boolean pool.active true
boolean pool.default true
int pool.importance 1
string pool.comment
pset pset_default

pset pset_default
int pset.sys_id -1
boolean pset.default true
uint pset.min 1
uint pset.max 65536
string pset.units population
uint pset.load 18
uint pset.size 1
string pset.comment

cpu
int cpu.sys_id 0
string cpu.comment
string cpu.status on-line

    Create the /etc/pooladm.conf file, pooladm -s save the active pool config to /etc/pooladm.conf

    # pooladm -s
# ls -ld /etc/pooladm.conf
-rw-r--r-- 1 root root 1007 May 3 14:59 /etc/pooladm.conf

    Create a pool set, called small, with at least one CPU with a max of 2 CPUs.

    # poolcfg -c 'create pset small (uint pset.min=1;uint pset.max=2)'

    Create a pool for the webservd user processes

    # poolcfg -c 'create pool webservd'

    Associate the apache pool with the small "pool set"

    # poolcfg -c 'associate pool webservd (pset small)'

    Instantiate the pool using

    # pooladm -c

    Adding resources which are not physically available on the system, you'll get the following.

    # pooladm -c
pooladm: configuration at '/etc/pooladm.conf' cannot be instantiated on current system

    Switch back to default pool

    # poolcfg -c 'create pool webservd'
# poolcfg -c 'associate pool webservd (pset pset_default)'
# pooladm -c

    List configuration

    # poolcfg -dc info

system default
string system.comment
int system.version 1
boolean system.bind-default true
string system.poold.objectives wt-load

pool webservd
int pool.sys_id 1
boolean pool.active true
boolean pool.default false
int pool.importance 1
string pool.comment
pset pset_default

pool pool_default
int pool.sys_id 0
boolean pool.active true
boolean pool.default true
int pool.importance 1
string pool.comment
pset pset_default

pset pset_default
int pset.sys_id -1
boolean pset.default true
uint pset.min 1
uint pset.max 65536
string pset.units population
uint pset.load 18
uint pset.size 1
string pset.comment

cpu
int cpu.sys_id 0
string cpu.comment
string cpu.status on-line

    Move project to the new pool.

    # projmod -sK "project.pool=webservd" group.webservd

    Bind to the apache pool

    # poolbind -p webservd -i project group.webservd

    This is how the /etc/project file looks now.

    # cat /etc/project
system:0::::
user.root:1::::
noproject:2::::
default:3::::
group.staff:10::::
group.webservd:102:Apache Default:::project.cpu-shares=(privileged,20,none);project.
max-shm-memory=(privileged,209715200,deny);project.pool=webservd

    Add utilization threshold and check.

    # poolcfg -dc 'modify pset pset_default (string pset.poold.objectives="utilization<75")'
# poolcfg -dc info

system default
string system.comment
int system.version 1
boolean system.bind-default true
string system.poold.objectives wt-load

pool webservd
int pool.sys_id 1
boolean pool.active true
boolean pool.default false
int pool.importance 1
string pool.comment
pset pset_default

pool pool_default
int pool.sys_id 0
boolean pool.active true
boolean pool.default true
int pool.importance 1
string pool.comment
pset pset_default

pset pset_default
int pset.sys_id -1
boolean pset.default true
uint pset.min 1
uint pset.max 65536
string pset.units population
uint pset.load 21
uint pset.size 1
string pset.comment
string pset.poold.objectives utilization<75

cpu
int cpu.sys_id 0
string cpu.comment
string cpu.status on-line

    Bind pool to the current process

    $ su - webservd
$ /usr/sbin/poolbind -q $$
11714 webservd

    Add Load to the system, to test threshold

    $ /usr/sbin/nspinx -n 4

    Check logs.

    # tail -f /var/logs/pool/poold
Aug 17 12:01:27 Monitoring INFO: pset pset_default utilization objective
not satisfied (1, utilization, '<', 75) with utilization 100.00 (control zone bounds exceeded)
Aug 17 12:01:27 Monitoring INFO: reconfiguration required
Aug 17 12:01:27 Optimization INFO: no moves found
Aug 17 12:01:42 Monitoring INFO: pset pset_default utilization objective
not satisfied (1, utilization, '<', 75) with utilization 100.00 (control zone bounds exceeded)
solaris 265
Log in to reply
 

© Lightnetics 2024