Re: un-import a SMF manifest

on solaris 10, manifest can be imported in one of two ways:
1. svccfg import /path/to/manifest.xml
or
2. imported automatically during boot if /var/svc/manifest
via svc:/system/manifest-import
now Is there a way to un-import a smf manifest ? what is the
prefered way to remove a manifest and associated stuff?
Thanks in advance.
Jason

Great but what if I get this:
svcadm disable tomcat
svcadm: Pattern 'tomcat' matches multiple instances:
svc:/application/servers/tomcat:default
svc:/network/tomcat:default
in order to delete it you need to disable it
first...how do I make svc look for a diferrence ?You are providing an abbreviation ('tomcat'). Svcadm is complaining that your abbreviation is not unique. Provide the entire service name (or at least enough so that it is unique).
Darren

Similar Messages

Availability of SMF Manifest for SRSS3.1

Does anyone know whether there is a SMF manifest for SRSS 3.1 available?

SRSS has not yet been adapted to run under SMF. It delivers no manifests.

SMF Manifests for Apache & MySQL

I recently started checking up on SMF since the need for service control became bigger. That and the options for added security when it comes to the privilege system. I dug up a manifest for apache somewhere but simply disliked it because it used the root account. So one thing led to another and you can understand the rest yourself.
This works for me, adds a little extra security, and I hope it also works for some of you. I tend to dump these in /var/svc/manifest/application/local and as you can see they all reside in svc:/application/local as well. Naturally this boils down to personal taste.
apache.xml manifest
<?xml version="1.0"?>
<service_bundle type='manifest' name='Apache'>
<service name='application/local/apache' type='service' version='1'>
<single_instance />
<dependency
name='filesystem_and_network'
grouping='require_all'
restart_on='none'
type='service'>
<service_fmri value='svc:/milestone/local-env' />
</dependency>
<dependency
name='configuration'
grouping='require_all'
restart_on='none'
type='path'>
<service_fmri value='file://localhost/etc/apache/httpd.conf' />
</dependency>
<exec_method
type='method'
name='start'
exec='/usr/apache/bin/apachectl start'
timeout_seconds='60' >
<method_context>
<method_credential
user='webservd'
group='webservd'
privileges='basic,net_privaddr,!proc_info'
limit_privileges='basic,net_privaddr,proc_chroot,proc_setid' />
</method_context>
</exec_method>
<exec_method
type='method'
name='stop'
exec='/usr/apache/bin/apachectl stop'
timeout_seconds='60' />
<exec_method
type='method'
name='restart'
exec='/usr/apache/bin/apachectl restart'
timeout_seconds='60' />
<instance name='default' enabled='false' />
<stability value='Evolving' />
</service>
</service_bundle>
--= EOT =--
And now for MySQL (note; this doesn't work for 5.x, I already have a test version but it still needs some work).
mysql.xml manifest
<?xml version="1.0"?>

<service_bundle type='manifest' name='MySQL'>
<service name='application/local/mysql' type='service' version='1'>
<single_instance />
<dependency
name='filesystem'
grouping='require_all'
restart_on='none'
type='service'>
<service_fmri value='svc:/milestone/local-env' />
</dependency>
<dependency
name='configuration'
grouping='require_all'
restart_on='none'
type='path'>
<service_fmri value='file://localhost/etc/my.cnf' />
</dependency>
<exec_method
type='method'
name='start'
exec='/etc/sfw/mysql/mysql.server start'
timeout_seconds='120' >
<method_context>
<method_credential
user='mysql'
group='mysql'
privileges='basic,!file_link_any,!proc_info,!proc_session'
limit_privileges='basic,proc_chroot,proc_setid' />
</method_context>
</exec_method>
<exec_method
type='method'
name='stop'
exec='/etc/sfw/mysql/mysql.server stop'
timeout_seconds='120' />
<instance name='default' enabled='false' />
<stability value='Unstable' />
<template>
<common_name>
<loctext xml:lang='C'>MySQL RDBMS 4.0.24</loctext>
</common_name>
<documentation>
<manpage title='mysql' section='1' manpath='/usr/sfw/share/man' />
</documentation>
</template>
</service>
</service_bundle>
--= EOT =--
As you can notice both of them depend on the milestone/local-env. This is my own creation. Because I'm using /var/run/apache and /var/run/mysqld I setup a shellscript to create them and linked this to a milestone entry. You can just leave it out or create your own manifest. I wonder if this forum can hold a message this long :-)

Well, I just realized that without the milestone manifest you're missing out on some essential dependencies (network & filesystem) so I just dump it into 2 messages 8-)
local-env.xml milestone
<?xml version="1.0"?>

<service_bundle type='manifest' name='Local_environment'>
<service name='milestone/local-env' type='service' version='1'>
<create_default_instance enabled='false' />
<single_instance />
<dependency
name='filesystem'
grouping='require_all'
restart_on='error'
type='service'>
<service_fmri value='svc:/system/filesystem/local' />
</dependency>
<dependency
name='network'
grouping='require_all'
restart_on='none'
type='service'>
<service_fmri value='svc:/milestone/network' />
</dependency>
<exec_method
type='method'
name='start'
exec='/usr/local/sbin/setlocenv start'
timeout_seconds='5' />
<exec_method
type='method'
name='stop'
exec=':false'
timeout_seconds='0' />
<property_group name='general' type='framework'>
<propval name='duration' type='astring' value='transient' />
</property_group>
<stability value='Evolving' />
<template>
<common_name>
<loctext xml:lang='C'>
Local configuration milestone
</loctext>
</common_name>
</template>
</service>
</service_bundle>
--= EOT =--
Keep in mind that this may work or simply won't do what you want. For example; whenever a filesystem error occurs the milestone will restart (thus re-creating /var/run/mysql and such and setting the proper ownership) but the real services (apache,mysql) won't respond to this. This maybe unsuited behaviour for you, in that case
just readjust the manifest to better suit your liking :)
You'll also notice that squid is mentioned, thats another project but although things work I'm not happy with it yet. Still; doesn't matter for the manifest since it basicly calls a shellscript which does the actual creating/setting of the proper directories (and even that need depends on how you configured mysql/apache).
Hope this works for you.

SMF manifest go to maintenace mode

I wrote the very simple manifest file to set 'ndd' parameter like
ndd -set /dev/tcp tcp_smallest_anon_port 42767
SMF service 'ndd' I created go to maintenace mode, I could not go online mode. how can I go to online mode?
Thank you for your help.
-Yong
# svcs ndd
STATE STIME FMRI
maintenance 19:21:20 svc:/network/ndd:default
Below is my manifest file for 'ndd' service.
<?xml version='1.0'?>
<!DOCTYPE service_bundle SYSTEM '/usr/share/lib/xml/dtd/service_bundle.dtd.1'>

<service_bundle type='manifest' name='ndd'>
<service
name='network/ndd'
type='service'
version='1'>
<exec_method
type='method'
name='start'
exec='/lib/svc/method/nddss %m'
timeout_seconds='10'>
<method_context>
<method_credential user='root' group='root' />
</method_context>
</exec_method>
<exec_method
type='method'
name='restart'
exec='/lib/svc/method/nddss %m'
timeout_seconds='10'>
<method_context>
<method_credential user='root' group='root' />
</method_context>
</exec_method>
<instance name='default' enabled='true'/>
</service>
</service_bundle>

Try adding
<dependency
         name='fs'
         grouping='require_all'
         restart_on='none'
         type='service'>
          <service_fmri value='svc:/system/filesystem/minimal' />
     </dependency>
     <dependency
         name='net'
         grouping='require_all'
         restart_on='none'
         type='service'>
          <service_fmri value='svc:/network/initial' />
     </dependency>

Why no SMF manifest instance after installing the apache2 packages?

Hi,
After installing the apache2 packages:
# pkgadd -G -d . SUNWapch2r SUNWapch2u SUNWapch2d
then type:
# svcs -a | grep apache2
the output is:
svcs: Pattern 'apache2' doesn't match any instances
STATE STIME FMRI
How do I solve it?
Thanks,
Terence

I am seeing the same problem. After installing the apache2 pkgs ...
svcs -l apache2
fmri svc:/network/http:apache2
name Apache 2 HTTP server
enabled true
state maintenance
next_state none
state_time Tue Apr 28 15:41:45 2009
logfile /var/svc/log/network-http:apache2.log
restarter svc:/system/svc/restarter:default
contract_id
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)
bash-3.00# tail -10 /var/svc/log/network-http:apache2.log
[ Apr 28 14:50:32 Executing start method ("/lib/svc/method/http-apache2 start") ]
[ Apr 28 14:50:32 Method "start" exited with status 96 ]
[ Apr 28 15:24:42 Leaving maintenance because disable requested. ]
[ Apr 28 15:24:42 Disabled. ]
[ Apr 28 15:28:06 Executing start method ("/lib/svc/method/http-apache2 start") ]
[ Apr 28 15:28:06 Method "start" exited with status 96 ]
[ Apr 28 15:38:22 Leaving maintenance because disable requested. ]
[ Apr 28 15:38:23 Disabled. ]
[ Apr 28 15:41:45 Executing start method ("/lib/svc/method/http-apache2 start") ]
[ Apr 28 15:41:45 Method "start" exited with status 96 ]
bash-3.00# svcs -x apache2
svc:/network/http:apache2 (Apache 2 HTTP server)
State: maintenance since Tue Apr 28 15:41:45 2009
Reason: Start method exited with $SMF_EXIT_ERR_CONFIG.
See: http://sun.com/msg/SMF-8000-KS
See: httpd(8)
See: /var/svc/log/network-http:apache2.log
Impact: This service is not running.
I have yet to find Status [96]

Checking SMF manifest exec_method 'start' : why does it execute multiple

I have a shell script given here in exec='/opt/hsqrt/bin/hsqrtctl'
q1: should that script exit with 0 immediately after process is started or
should it wait until process dies or gets killed ?
q2: where do I find further information hwo to write a start & stop exec_method ?
q3: how do I avoid a second start while my process (rtserver) is still running ?
<exec_method
type='method'
name='start'
exec='/opt/hsqrt/bin/hsqrtctl start -id all'
timeout_seconds='15' >
</exec_method>
<exec_method
type='method'
name='stop'
exec='/opt/hsqrt/bin/hsqrtctl stop -id all'
timeout_seconds='15' >
</exec_method>

The script should exit with 0 as soon as the service is ready to serve requests.
The best documentation for service authors we have right now is at http://www.sun.com/bigadmin/content/selfheal/sdev_intro.html . Otherwise you can send questions to [email protected] , and look at its archives.
After your start method exits with 0, svc.startd should mark your service as "online", and then it won't execute your start method even if the user runs "svcadm enable" again.
David

SMF - start at boot time, but don't restart - possible?

I'm setting up a software package that has a number of background processes.
I've got SMF manifests created and imported for all of them, and they're working great for starting the processes at boot time, and doing the automatic restarts when one fails.
Here's my problem. It turns out that the users of this software want to be able to use the software's integrated process control methods to stop some of their processes. With the current configuration, they use the software's method to stop one of it's processes, and SMF promptly restarts it.
Is there a way to set up SMF managed software so that SMF starts it at boot time, but then doesn't care if stops later?
Richard

I'm not sure how it would be SMF "managed" in such a case. Obviously you want the management to be elsewhere.
You could use the "legacy" startup scripts in /etc/rcX.d. That will start a process and forget about it, just like pre-Solaris 10. You don't use manifests or have any other benefits of SMF.
If you do want SMF, I guess the question is how you want it to act..
When the users stop the processes, do you want SMF to say enabled or disabled? SMF does not track enabled processes by periodic polling, it does it by tracking contracts. This means that starting a process outside of SMF does not allow SMF to track it.
Darren

Assistance with Error - virtualmachinemanager.psd1 module cannot be imported, one or more members are not valid.

Hello,
We use Orchestrator, System Center 2012 version 7.1.3022.0. I have an orchestrator script that reads through a txt file to find unresponsive virtual machines (Windows 2000 VM's cannot have the SCOM agent installed on them) so we can restart any
machines that are not responding.
The step to Get Computer/IP Status runs successfully in that it reports which VM's are running and which are not running. The next step then is to run a Powershell script that restart the VM's that are down. Here is my Powershell Script:
$ErrorActionPreference = "Stop"
try
Import-Module 'C:\Program Files\Microsoft System Center 2012\Virtual Machine Manager\bin\psmodules\virtualmachinemanager'
Get-VMMServer -ComputerName mlivsmsscvmm
# get vm object
$vm = get-SCVirtualMachine | where { $_.ComputerName -eq "\`d.T.~Ed/{A255D4D8-BD4D-4FF4-9A51-B1ABA82720B9}.Computer-name\`d.T.~Ed/" }
# save data to publish for future activities
$OrigVMMStatus = $vm.StatusString;
$AlertAdded = "\`d.T.~Ed/{A4B4EC8A-E4F8-40AB-9897-6906EAE139AE}.TimeAdded\`d.T.~Ed/";
$VMName = $vm.Name;
# we are at this point because the VM is not responding.
# Start VM - if VM status is "Running" then Stop the VM first
if($vm.StatusString -eq "Running")
{ Stop-SCVirtualMachine -vm $vm -force };
Start-Sleep -s 20;
if($vm.StatusString -eq "Stopped")
{ Start-SCVirtualMachine -vm $vm
$VMrestart = "Yes"
catch
Throw $_.Exception
Upon running the Powershell script I receive the following error in Orchestrator Log histry:
The 'C:\Program Files\Microsoft System Center 2012\Virtual Machine Manager\bin\psmodules\virtualmachinemanager\virtualmachinemanager.psd1' module cannot be
imported because its manifest contains one or more members that are not valid.
This script used to run without errors and nothing has been upgraded in our environment with regards to Orchestrator, VMM, SCOM. How would I find the 'one or more members' that is causing my issue?
Thank you.

Hi,
the thing is that PowerhShell in the "Run .Net Activity" is version 2.0 (http://www.sc-orchestrator.eu/index.php/scoblog/102-running-powershell-30-with-the-run-net-script-activity)
So, try your script embedded in PowerShell {}
Regards,
Stefan
www.sc-orchestrator.eu ,
Blog sc-orchestrator.eu

Standard Midi File Import/Export Questions

Hi,
I'm having some trouble with Standard MIDI files, in that, after exporting them, and importing them back into another Logic project to test them, some of them will split/demix. That is, instead of one MIDI region appearing, it imports as two MIDI regions, with the notes of the part split between the two regions (the notes seem to split according to velocity, not note value).
Here is my question related to this:
Is it possible to export a Standard MIDI File for each "part" in a project, rather than one SMF for all the MIDI files in a project? (i.e., create one SMF for each software instrument track)? This is what I'm trying to do. The reason is that I'm trying to prepare and save some MIDI files to include with the audio files in a remix pack, and, since it's likely that the various remixers will be using different DAWs such as Cubase, I'm assuming that a SMF that includes all the various parts might not import into their program correctly. Is my assumption wrong? I.E., will most DAWs, like Cubase, etc., correctly import a SMF that includes all parts, and upon import split those parts correctly?
I'm following all the instructions for "Saving Standard MIDI Files" from the Logic Pro manual, except that I am not merging all the different instruments' MIDI regions together; I'm only merging each individual parts' MIDI regions.
Thanks!
Message was edited by: trobert225 - I accidentally posted before I finished the question.

Well I think the problem is solved. There was a conflict between Logic and iLife 06. I bought a new computer this week and had to load everything into it. I decided not to put in iLife 06 yet including Garageband and all the jam packs. I loaded all the new .caf files off the Logic DVD
I opened a midi file and all tracks loaded with Logic instruments. the problem was that a lot of the instruments had been erased somehow or over written by Logic studio. Now I believe they are all there and things are working as they should

Question related smf service config

Hello,
I'm trying to create some smf manifests which should start some java vm which represents a whole application. There are for example a vm which is the jndi, a vm which works as jms and a "first" app which is called "manager".
So I created a manifest "manager.xml" and a second manifest called "jms.xml". The jms.xml includes a dependency to manager.xml - jms should be dependent from manager:
<dependency name='manager'
                grouping='require_all'
                restart_on='refresh'
                type='service'>
        <service_fmri value='svc:/application/barisk/manager'/>
</dependency>
In the manager.xml is a section dependent placed:
<dependent name='jms'
                grouping='optional_all'
                restart_on='none'
                type='service'>
        <service_fmri value='svc:/application/barisk/jms'/>
</dependent>
My problem is now: if the manager and the jms are disabled and I start the jms with "svcadm enable jms" how it's possible to trigger the smf to enable and start the manager too without starting the manager with "svcadm enable manager" manually?
After starting the jms the state is:
svc:/application/barisk/jms:default (Product Serial Number Collector)
State: offline since Tue Jul 30 10:01:37 2013
Reason: Service svc:/application/barisk/manager:default is disabled.
   See: http://sun.com/msg/SMF-8000-GE
Path: svc:/application/barisk/jms:default
          svc:/application/barisk/manager:default
Impact: This service is not running.
Thank you for any suggestions,
Martin

Ok, thank you,
in the meanwhile I understand the basic concept of smf. But at first I was surprised that enabling the manager is not the job of smf.

Process dumps core, when started by SMF - fine when manually started

Hi.
h2. Explanation
I compiled [Nagios 3.2.0|http://www.nagios.org/] manually, and created a SMF manifest and method file, based on the [OpenCSW supplied files for Nagios|http://www.opencsw.org/packages/nagios]. The method file is based on [what the Nagios guys supply|http://pastebin.ca/1577792] , but [changed a little|http://pastebin.ca/diff.php?id1=1577792&id2=1577793] .
When I enable the "application/nagios" service, it right away goes into maintenance state. In the /var/svc/log/application-nagios:default.log log file, there's consequently then this:
+ exit 0
[ Sep 24 13:17:10 Method "start" exited with status 0 ]
[ Sep 24 13:17:11 Stopping because process dumped core. ]
[ Sep 24 13:17:11 Executing stop method ("/opt/local/nagios-3.2.0/method/nagios stop") ]
+ test -r /lib/svc/share/smf_include.sh
[...]It died, because of "*Stopping because process dumped core.*"
If you have a look at the method file, you'll find that I dump the environment with env to a file. When I login as root and set the environment variables to exactly those values and remove any other additional variables, I can run the method / init script just fine; the nagios process doesn't core dump.
h2. Question
Why does the process dump core, when it's run by SMF?
Additional questoin:Where's the core dump? I did a "find / -name core" and found nothing. I also changed coreadm:
--($ ~)-- LC_ALL=C coreadm
     global core file pattern: /var/core/core.%f.%pAfter a reboot, there's nothing in /var/core (the directory exists).
I'm on Solaris 10 5/09 Sparc.
h2. Files
*[svccfg export nagios|http://pastebin.ca/1577787]*
*[Method file /opt/local/nagios-3.2.0/method/nagios|http://pastebin.ca/1577793]*

There's no such directory as "/var/crash":
--($ ~)-- ls -lad /var/c*
drwxr-xr-x    4 root     sys             4 Sep 21 14:03 /var/cache
drwxr-xr-x    2 root     sys             7 Sep 21 13:32 /var/cc-ccr
drwxr-xr-x    2 root     root            2 Sep 24 13:14 /var/core
drwxr-xr-x    2 root     sys             3 Sep 21 14:03 /var/cronGranted, since when I posted that question and now, I booted the NGZ numerous times. But in a default installation of Solaris 10, that directory isn't removed during shutdown/boot, if it exist. At least it exists on my other servers, even if it is empty (eg. like that, on a different server:
--($ ~)-- ls -laR /var/crash/
/var/crash/:
total 3
drwxr-xr-x    3 root     root          512 Mai 19 2006 .
drwxr-xr-x   56 root     sys          1024 Dez 7 2008 ..
drwxrwx---    2 root     adm           512 Sep 22 2008 winds06
/var/crash/winds06:
total 2
drwxrwx---    2 root     adm           512 Sep 22 2008 .
drwxr-xr-x    3 root     root          512 Mai 19 2006 ..So, I'd say that the coredump has never been stored in /var/crash.
Anyway - isn't /var/crash just used for crashdumps, when the whole machine goes down? According to [man filesystem(5)|http://docs.sun.com/app/docs/doc/816-5175/filesystem-5?a=view] , it is → "+Default depository for kernel crash dumps.+"
If coredumps aren't in /var/crash, aren't named "core*" and aren't stored in the directory specified by coreadm, where are they?
Converse: If no coredump is created, why does Solaris "speak" about a coredump being created?

Jumpstart , smf and scripts on first boot

Could anyone help me I am trying to work out how which is the best way to apply patches and customizations on first boot of a new machine that has just been jumpstarted. Does anyone have a example of a service that runs once to allow me to do this?

I want to be able to not just add patches I need to
add packages make system changes I used to be able to
do this with /etc/rc2.d scripts but now its smf in
solaris 10 and I need to be able to make this changes
at run level 2 and then reboot.You can do it just like before. Put a script in /etc/rc2.d.
You can also create a SMF manifest, and a custom "site" profile to activate/deactivate the services you want. Put them into the manifest and profile directories and they'll be loaded and activated/deactivated at first boot.
Darren

How to create a new bundle of service and import it as normal user.

Hi,
I have created a service.
Now am able to import and enable the service.
I have placed my xml file under
/var/svc/manifest/site
The service name is
<service
name='site/newservice'
I have imported the service as root user.
<exec_method type='method' name='start' exec='/tmp/new/method/newservice start' timeout_seconds='30' />
This service is not able to execute the operation which I have mentioned above. The following error is thrown in log file.
Error: User 'root' is not the owner of '/tmp/new/method'
So I tried to import the service as normal user, the following error is thrown.
svccfg import /var/svc/manifest/site/newservice.xml
svccfg: Could not create temporary service "TEMP/site/newservice" (permission de nied).
svccfg: Import of /var/svc/manifest/site/newservice.xml failed. Progress:
svccfg: Service "site/newservice": not reached.
svccfg: Instance "newservice": not reached.
Please help me to solve this issue.
I thought instead of placing the service in default location,Why cant we create a new service of bundle and place the service there.
Please help to solve this issue by suggesting some solution or how to create a new bundle of service and import it as normal user.
Regards,
Kalai

kalaiyarasan_P wrote:
Hi,
I have created a service.
Now am able to import and enable the service.
I have placed my xml file under
/var/svc/manifest/site
The service name is
<service
name='site/newservice'
I have imported the service as root user.Okay.
<exec_method type='method' name='start' exec='/tmp/new/method/newservice start' timeout_seconds='30' />
This service is not able to execute the operation which I have mentioned above. The following error is thrown in log file.
Error: User 'root' is not the owner of '/tmp/new/method'By default, services run as the 'root' user. It's complaining that the method isn't also root.
So I tried to import the service as normal user, the following error is thrown.You won't be able to do that easily. The user that does the import is almost always root. But wihin the manifest you could have it run as a different user.
I thought instead of placing the service in default location,Why cant we create a new service of bundle and place the service there.
Please help to solve this issue by suggesting some solution or how to create a new bundle of service and import it as normal user.You don't need to import this as a normal user, do you? You just want it to run as one, right?
Darren

Editing and Saving An SMF through Logic Express 8

Rather than importing an SMF into Logic Express I dropped the file in an opened it that way.After editing the track sounds etc. I tried exporting it but it only exports back into its original form,none of the changes made save to the exported SMF,does anybody know how to export the edited SMF rather than the original.Cheers,Damien

I'm running 10.6.5 and LE 8 and it works for me. Of course I have a new imac that came installed with 10.6.5 and I installed LE on it. Maybe you can re-install LE. But hopefully there is another solution. I too don't really want to upgrade to LE 9 if I don't have to.

Sun Cluster and SMF

Folks,
Did you come across any products, who SMFied their product and how it impacts their cluster installations ?. Please provide details if you have any
Thanks mUch
_D

Since I am currently not aware of third party products shipping an SMF service for their application (which does not mean much, ie. they could well exist) I just make two generic statements:
1) If for that application a standard Sun Cluster agent exists, then the instructions for installing this data service would include to disable the SMF on the cluster nodes (like on pre Solaris 10 systems the legacy runlevel script would need to get deactivated).
2) If for that application no standard Sun Cluster agent exists, then one would have the ability to leverage the new proxy SMF resource types coming with Sun Cluster 3.2. They would use the SMF manifest to start and stop the application. Note that by using this resource type you would not have any application fault monitoring.
Details to 2) can be found at
http://blogs.sun.com/SC/entry/making_smf_services_highly_available and
http://docs.sun.com/app/docs/doc/819-2974/6n57pdk2b?a=view
Greets
Thorsten
Message was edited by:
Thorsten.Frueauf

Re: un-import a SMF manifest

Similar Messages

Maybe you are looking for