Stacking simple jobs with batch on Solaris 10 x86
Hey Guys,
I have a user that needs to load up a bunch of very low priority (nice 19) jobs on a 4 CPU machine so that once one completes, another one starts. I thought batch would be a good use for this. We don't have any mpi worries or anything, these are very monolithic standalone jobs.
I looked at the batch man page and it seems to imply that I can create a NEW queue in /etc/cron.d/queuedefs to accomplish this. However, I don't see how to format it such that many jobs can be "loaded" in a new queue.
Here is the default:
bash-3.00# cat /etc/cron.d/queuedefs
a.4j1n
b.2j2n90w
bash-3.00#Should I modify the b (batch) line to say:
b50j19n90wBut it seems like the above would simply allow 50 jobs to be started at a nice level of 19 - is that correct?? What I don't see is how to specify that only 4 jobs should be run at one time (4 CPUs) but please accept ~50 jobs into a queue and load them as sequentially as jobs finish.
Should a make a new queue to be listed after the "b" queue in queuedefs?
Can anyone comment on this? Is this out of scope for batch and queuedefs ?
Thanks,
Phil
One way to do this would be by using the shell construct of &&.
command_one && command_two && etc...
&& will tell command two to wait for command one to finish with an exit status of zero before continuing on. If command one fails for any reason then the rest of the commands would not finish.
So to get around that you could use:
command one
sleep <time>
command two
sleep <time>
etc...
or:
man wait
might be the best option.
alan
Similar Messages
-
Help with evaluation for Solaris x86
Hello,
I would like to evaluate Oracle Identity Manager and Access Manager products at the latest version possible -- the host system(s) are Solaris x86 (SunOS .5.10) Unfortunately, I am having a bit of trouble compiling the list of packages/versions to achieve this for this platform. For instance, on the downloads page it appears that OIM is not available for Solaris x86, not even in older versions(?)
Thank you in advance,
ScottHi Scott,
I'm pretty sure that x86 Solaris is not yet certified to run the Identity Management suite. I've been involved in an evaluation process and we ran it all on RedHat Enterprise Server Release 4 Update 5, without too many problems.
cheers,
James -
Please suggest the fastest way to install Solaris x86 PC 30 boxes
I've already tested one and it's successfully installed.
I have to install 30 boxes , all are same spec.
Please advise the fastest procedure to install.
Note : There is no sparc machine.Only problem with imaging a Solaris x86 box is that your machines will end up with the same hostid (which is pseudo-randomly generated at Solaris installation time for x86). This may, or may not, be a problem, depending on what commercial software you use...
To avoid this, you have to use jumpstart, or do some serious hacking. -
Solaris x86 with Oracle RAC 10g Enterprise Edition Release 10.2.0.3.0
Hello,
Maybe you can help me (new on RMAN backup) in doing this.
I have configured a single Oracle 10g database to have backup with RMAN with following steps:
1. $ mkdir $ORACLE_BASE/rman_scripts
2. $ mkdir $ORACLE_BASE/logs
3. $ mkdir $ORACLE_BASE/tracking
4. $ mkdir $ORACLE_BASE/c_backup
5. $ sqlplus sys/<password> as sysdba
6. SQL> alter system set db_recovery_file_dest_size = 50G scope=both;
7. SQL> alter system set db_recovery_file_dest='${ ORACLE_BASE}/flash_recovery_ area' scope=both;
8. SQL> alter system set log_archive_dest_10='location= use_db_recovery_file_dest';
9. SQL> shutdown immediate
10. SQL> startup nomount
11. SQL> alter database archivelog;
12. SQL> alter database open;
13. SQL> alter database enable block change tracking using file '${ORACLE_BASE}/tracking/rman_ change_track.f';
14. $ rman target /
15. RMAN> CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK
TO '/var/opt/oracle/flash_ recovery_area/ORCL/c_backup/% F';
16. RMAN> CONFIGURE CONTROLFILE AUTOBACKUP ON;
17. RMAN> CONFIGURE BACKUP OPTIMIZATION ON;
18. RMAN> CONFIGURE RETENTION POLICY TO RECOVERY WINDOW OF 7 DAYS;
19. RMAN> exit
I need to configure incremental backup with RMAN on a two node Solaris x86 with Oracle RAC 10g Enterprise Edition Release 10.2.0.3.0 installation.
We also use ASM to store database files, and have Oracle software installed on separate file systems (two Oracle roots for Node1 and Node2).
I have following questions:
1) where to put Flash Recovery Area (FRA)?
I saw recommendations to put FRA on the ASM, is this the best way to do it?
2) Can I put FRA on another file system (not on the ASM) which is available only from Node1? This way I can save space on the ASM.
3) Is it possible/recommended to run RMAN from Node1 only?
Below is the script used to run RMAN on the normal Oracle database (without RAC) which I need to change :
=============================================================================================
2.0 Oracle backup script: /opt/app/oracle/rman_scripts/backup.sh
Use this for daily backups, possiblly as a cron job.
Once a week run this: /opt/app/oracle/rman_scripts/backup.sh FULL
All other days of the week: /opt/app/oracle/rman_scripts/backup.sh INCREMENTAL
Note: You may have to change ORACLE_SID, ORACLE_BASE below to match your database.
=============================================================================================
#!/usr/bin/ksh
ORACLE_SID=orcl
ORACLE_BASE=/opt/app/oracle
ORACLE_HOME=${ORACLE_BASE}/product/10.2.0/db_1
PATH=${ORACLE_HOME}/bin:/usr/bin
LOGDIR=${ORACLE_BASE}/logs
LOGFILE=${LOGDIR}/rman.log
if [[ $# < 1 ]]
then
echo "usage: backup.sh FULL|INCREMENTAL"
exit;
fi
BACKUPTYPE=${1}
full='FULL'
incremental='INCREMENTAL'
if [[ $BACKUPTYPE == $full ]]
then
$ORACLE_HOME/bin/rman target / nocatalog log ${LOGFILE} append << eof
run {
backup database;
SQL 'alter system archive log current';
backup archivelog all;
delete noprompt obsolete;
exit;
eof
echo ''
fi
if [[ $BACKUPTYPE == $incremental ]]
then
$ORACLE_HOME/bin/rman target / nocatalog log ${LOGFILE} append << eof
run {
backup database;
backup incremental level 1 database;
SQL 'alter system archive log current';
backup archivelog all;
delete noprompt obsolete;
exit;
eof
echo ''
fiHi [email protected],
Q1) where to put Flash Recovery Area (FRA)?
A1) With RAC: on the shared storage
I saw recommendations to put FRA on the ASM, is this the best way to do it?
If you want your backups to be available for both nodes you have to use shared storage or tape using an mml library.
So if you want to use the FRA for rman backups and the database is on ASM just make ASM the standard for the FRA as well.
Q2) Can I put FRA on another file system (not on the ASM) which is available only from Node1? This way I can save space on the ASM.
A2) Than you cannot recover in case Node1 is down. Best would be to send your storage admin to a training course so he can manage the clustered raw devices needed for ASM.
Q3) Is it possible/recommended to run RMAN from Node1 only?
A3) No see A2.
Regards,
Tycho -
Hi,
We have a program that will create a batch job and then we need to go to SM35 to run that batch job manually. We want to avoid this manual step and what we want to do is to as soon as our program runs it will create a batch job and then batch job should kick off automatically.
One of the solution that came to my mind is to schedule a batch job with two conditions:
1. in first condition it will run our main program
2. and in 2nd step it will look out for the batch job that created in SM35 and will kick off that, but I am not sure how can i do this using Sm36.
I was also wondering if there is another way so that we can just avoid the batch run for the second step?
So can you please help me with this.
Thanks,
RajeevSM35 -> batch sessions, so I assume you are talking about these.
You can start sessions under program control by submitting program RSBDCSUB after creation of the session.
Thomas -
Hi,
Is it possible to create a batch job with another user name using JOB_OPEN?
and what is the use of JOB_GROUP option??
Thx in advance.
PRause:
CALL FUNCTION 'JOB_OPEN'...
*--create Step
SUBMIT RFBIBL00 AND RETURN
USER SY-UNAME VIA JOB JOBNAME NUMBER JOBCOUNT
WITH DS_NAME = FILE
WITH CALLMODE = 'B'.
greetings -
Solaris x86 with card on-board
Hi all,
I got a problem when installing Solaris9 x86 on ibm NetVista Machine.
I have installed 2 other machines with separated network card sucessfully. I can configure to connecto to my LAN network.
But when install on 2 ibm NetVista machines (which include onboard network card ), It is fail to be configured to connect to the network in both following cases:
Case 1: I choose proper driver as Sun suggest for Intel VE Pro(R) 10/100 : iprb. And the Solaris detect successfully. I checked by "ifconfig -a" and saw the network card active on "iprb0".
Next, I following correct instructions to configure the network card to connect to the LAN. But when i ping IP (9.187.186.18) to other computer in our same subnet LAN,
The system display eg: "No anwer from 9.187.186.18"
I run the "snoop" command to view broadcasr ARP in an other terminal. The snoop terminal displays ARP signal boradcasting from the Solaris. However, it can nerver get ARPs from other computer in LAN.
Case 2: I add 3COM network driver to the solaris ( onboard card is no loger used ). The system can auto-detected this card and i can configure to connect to the network. The problem remain unchange.
Is there any special in IBM Netvista machine need to be configure under Solaris x86 or any particular configuarion for on-board network card. ???
Can any boy help me to solve this problem ??
Best regards,
Thang,
Vietsoftware DeveloperHowever, it can nerver get ARPs from other computer in LAN.Are you perhaps running "snoop" on a box different from the one that you're trying to ping, and are you using
a switched ethernet? In that case it would be OK if you only see the ARP broadcast, because the ping target
box will send the ARP reply directly to the MAC of the requesting station (i.e. there's no ARP broadcast reply),
and an ethernet switch will send the ARP reply directly to the port where the station with the given destination
MAC is connected - no other port should see the reply.
If you're trying to ping 9.187.186.18 from the new Solaris x86 box, try to run snoop on the 9.187.186.18
system. 9.187.186.18 should see the ARP broadcast from the LAN, and the 9.187.186.18 box should list the
ARP response it sends back to the LAN.
Is there any special in IBM Netvista machine need to be configure under Solaris x86 or any particular
configuarion for on-board network card. ???I guess the interrupt handler for the NIC is not working, so the Solaris x86 sees no incomming packets.
See the "troubleshooting" section on the following page:
http://solaris-x86.org/documents/tutorials/rtls.mhtml -
Help with oracle 10g install on solaris x86
hi, i downloaded solaris 10 from oracle site (sol-10-u8-ga-x86-dvd.iso) and i have installed it on vmware. i was going through the installation guide for oracle 10 for solaris x86 and while on the pre-installation steps, i issued this command
# /bin/isainfo -kv
and it gave me
# 64-bit amd64 kernel modules
instead of
# 32-bit i386 kernel modules
as the guide expected.
please what can i do, cos the guide says if i dont see the later, the installation cannot complete.
thnxFirst check you Operating System Bit information whether it is 64 or 32 bit , for this first check any of the command path
Example ls command
$ which ls
/usr/sbin/ls
$ file /usr/sbin/ls
It will show you whether your OS is 64 bit or 32 bit and accordingly download the software from below link ,if it is showing 64 bit os then you have to download 64 bit database software .
http://www.oracle.com/technetwork/database/database10g/downloads/index.html -
Installing Solaris 9 with other Operating Systems x86
Hi, I am new to Solaris world. I want to install solaris 9 on my home computer on which I've already installed Windows XP and RedHat Linux 9. I have a few questions regarding the installation process.
(a) Is a primary partition necessary for installing Solaris 9 (I have only one primary partition and free space in extended partition)? Is there a way to boot Solaris 9 from a logical drive in the extended partition?
(b) Will choosing the default option on the Solaris installation fdisk screen erase my previous partitions?
I will appreciate if anybody care to answer.hi sulemankm
solaris needs a primary partition, booting from a logical drive won't work.
for very detailed information take a look at
http://multiboot.solaris-x86.org/
hth
jenny -
SGD Solaris x86 no SecurID support
Hi,
Is there no support for RSA SecurID on Solaris x86. SGD version 4.2. It seems as if the RSA SecurID module is just not installed (with security pack!).
Regards,
Arno Staal
Divider B.V.Call or write RSA and let them know you want SecurID for Solaris X86. There is no excuse they don't release this as it is exactly the same code as Sparc, just a simple recompile.
I just spoke to them and they are checking on it for me.
Marc -
How about gccfss for Solaris x86/x64 platforms?
Subject says it all... it could make compiling a lot of code out there a whole lot easier!
BobI also desire gccfss for Solaris x86/x64 platforms .
In the interim I have downloaded the source and am attempting to create such a beast.
It is slow progress to fix the unexpected script breakage, but it seems to be going OK:
Problems:
1.): Issues like "-mtune=v9" and an "embeded spec" in gcc and cpp (from files
gccfss_src/build/gcc/gccspec.o and gccfss_src/build/gcc/cppspec.o).
2.): Use of OPTTABLE_H = $(srcdir)/config/target-option-table.h with SPARC specific
options throughout
EG:
#ifdef CROSS_COMPILE
#define NATIVE \
{ "-native", "-xtarget=generic" }, \
{ "-xarch=native64", "-Zarch=v9 -Zarchm64=v9 -Zm=64" },\
{ "-xarch=native", "-Zarch=v8plus -Zarchm32=v8plus -Zarchm64=v9 -Zm=32" },\
{ "-xtarget=native64", "-Zarch=generic -Zarchm32=generic -Zarchm64=v9 -xchip=generic -xcache=generic -Zm=64"}, \
{ "-xtarget=native", "-Zarch=generic -Zarchm32=generic -Zarchm64=v9 -xchip=generic -xcache=generic -Zm=32"}
#else
#define NATIVE \
{ "-native", "-xtarget=native" }, \
{ "-xarch=native64", "-xarch=native64 -Zm=64"}, \
{ "-xarch=native", "-xarch=native -Zm=32"}, \
{ "-xtarget=native64", "-xtarget=native64 -Zm=64"}, \
{ "-xtarget=native", "-xtarget=native -Zm=32"}
#endif
Thus, one would need to change:
{ "-xarch=native64", "-Zarch=v9 -Zarchm64=v9 -Zm=64" },\
to something like:
{ "-xarch=native64", "-xarch=386 -m64" },\
3.): Copy the "sparc_output_scratch_registers()" function from file
"gccfss_src/src/gcc/config/sparc/sparc.c" to "gccfss_src/src/gcc/tree-ir.c".
4.): Use the undocumented command line option "-frtl-backend" when this error
occurs during the build (a simple Makefile change):
`-mcpu=' is deprecated. Use `-mtune=' or '-march=' instead.
xgcc: SUNW0scgfss 4.2.0 has not been installed. Either install it or use -frtl-backend.
5.): More ...
All these issues occur because the scripts expect "--build=sparc-pc-solaris2.11" but
I configured using "--build=i386-pc-solaris2.11" . It seems like it is going to work
but I am yet to finish stage1 of the build (I do have a working "xgcc").
Rob -
Creating backup Job with DBMS_SCHEDULER
Hello,
Can someone please help me out here:
I'm using Oracle10g release 1 on windowsXP
I'm trying to create a backup job with dbms_scheduler and it's not working.
This is what I did:
I created a job as follows:
BEGIN
dbms_scheduler.create_job (
job_name => 'RMAN_FULL',
job_type => 'EXCUTABLE',
job_action => 'E:\wkdir\rman_bkp',
enabled => TRUE,
start_date => '24-NOV-2007 2:10:00 PM',
repeat_interval => 'FREQ=WEEKLY',
comments => 'Full Database Backup');
END;
While rman_bkp is an RMAN command but it wasn't working.
Please where do I get it wrong?
Kindly put me through the EXECUTABLE or should I use PL/SQL_BLOCK and how?
Thanks.
Regards,
CherishHi,
There is a guide to running external jobs using the Scheduler here
Guide to External Jobs on 10g with dbms_scheduler e.g. scripts,batch files
You need to use the full path to a real Windows executable and the arguments to it e.g. for a batch script you would have to do something like
c:\windows\cmd.exe /q /c c:\myscript.bat
There is a forum dedicated to the Scheduler here
Scheduler
Hope this helps,
Ravi. -
Hi,
This might sound really remedial, but can someone explain to me how I go about creating a job with arguments with programs. It errors out when I try to use DEFINE_PROGRAM_ARGUMENTS and when I try to use SET_JOB_ARGUMENTS_VALUE. Maybe it's the order I do it in. Any directions would be much appreciated. Thanks!
TonyHi Tony,
Almost any internal error is an Oracle bug and you should report this to support.
This is definitely not supposed to happen and I haven't seen this error before, do you have a test case throwing the error ?
Using program arguments should be fairly straightforward. Here's a simple example with two varchar2 arguments
-- create a stored procedure with two arguments
create or replace procedure myproc (arg1 in varchar2, arg2 in varchar2)
is BEGIN null; END;
-- create a program with two arguments and define both
begin
dbms_scheduler.create_program
program_name=>'myprog',
program_action=>'myproc',
program_type=>'STORED_PROCEDURE',
number_of_arguments=>2, enabled=>FALSE
dbms_scheduler.DEFINE_PROGRAM_ARGUMENT(
program_name=>'myprog',
argument_position=>1,
argument_type=>'VARCHAR2',
DEFAULT_VALUE=>'13');
dbms_scheduler.DEFINE_PROGRAM_ARGUMENT(
program_name=>'myprog',
argument_position=>2,
argument_type=>'VARCHAR2');
dbms_scheduler.enable('myprog');
end;
-- create a job pointing to a program and set both argument values
begin
dbms_scheduler.create_job('myjob',program_name=>'myprog');
dbms_scheduler.set_job_argument_value('myjob',1,'first arg');
dbms_scheduler.set_job_argument_value('myjob',2,'second arg');
dbms_scheduler.enable('myjob');
end;
Hope this helps,
Ravi.
-Ravi -
View jobs with SQL Developer and export to file with delimiter
Hi,
is there a way to view, create and alter my database jobs with SQL developer?
And is it possible to export tables, views or simple queries to file delimited different from the CSV standard?
Diego.Hi and Welcome to the Forum!!
To select data from a table or a view you must use the SELECT sql command. Example to get the data from your view you can
select ZMATNUM,ZMATCUR,ZMATATTR from ZMATERIALAs you have said you are using SQL Developer just execute the above query in your sql developer and it will show the data. Right click the grid and export the data to excel file. -
Oracle 8.1.5 Client/WLS/Solaris x86
has anyone been able to get Oracle's client 8.1.5 running on Solaris x86 in conjunction with WLS to talk to back-end Oracle? Any help would be greatly appreciated.thanks.
This will not work. We have not built the native libraries to work with
Solaris on Intel..
I suggest trying the platform independent type 4 JDBC driver available for
free from Oracle. It is supported (as is any JDBC driver) with WebLogic
Server. To download it:
Go to http://www.oracle.com and select the "Download" option.
From the resulting page, use the "Select Utility or Driver" dropdown to
select Oracle JDBC drivers
From the resulting page, scroll down a little (since SQLJ stuff appears at
the top).
Or, to go directly there:
http://technet.oracle.com/software/tech/java/sqlj_jdbc/software_index.htm
Thanks,
Michael
Michael Girdley
BEA Systems Inc
"Dean Davis" <[email protected]> wrote in message
news:3a267487$[email protected]..
has anyone been able to get Oracle's client 8.1.5 running on Solaris x86in conjunction with WLS to talk to back-end Oracle? Any help would be
greatly appreciated.thanks.
Maybe you are looking for
-
Now that I am on Mavericks and no longer able to sync between my iPhone/iPad and Outlook 2011 on my MacBook Pro, I'm finally taking the leap (which I should have done long ago) and abandoning Outlook 2011 all together (doesn't help that Outlook now p
-
Macbook pro not fully charging
My macbook pro is less than one month old and it started not fully charging a few days ago. It was at 97% when the light turned green, and wouldn't go up any more. I always have my keyboard light off & the screen light very dim so don't tell me to lo
-
2MEM/1/4:DIMM2/J13 Error code
My PowerMac recently reported this error during an apple hardware test. I have the factory 512 ram installed, plus additional 6gb of ram in 1gb cards. I still get this error message no matter which configuration of ram cards i use, e.g. if i move or
-
Exception handling within a value-binding expression
Hi all, Forgive me if this question seems odd, but I am a long-time Struts developer, new to JSF. In Struts, exception handling was easy because all requests were funnelled through some subclass of Action. Exception handling could effectively be cons
-
Is there a way to "catch" an integrity constraint in PL/SQL. the integrity constraint that i want to catch is a violation to the foreign key. the error number is ORA-OO291 Regard, Richard