Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Alessandro Perucchi

Pages: 1 ... 6 7 8 9 10 [11] 12 13 14 15 16 ... 65
151
MP Server / Re: ARSMAINT increase ARSSOCKD process count
« on: August 02, 2016, 03:36:32 AM »
Hi .
 
Just  grep process PID from the Shell process  and Kill.

ps -ef |grep arsmaint

kill -9 PID

it will stop the arsmaint process.


NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

If you kill a process PLEASE do you use the following command:

kill PID

AND ONLY if that command does not work, then use the kill -9 PID.

The -9 is really the option to use when everything was done to stop a process and it didn't work.
When doing a -9 you are facing a possible corruption of data, because the process was murdered without leaving him the time to close cleanly the transaction it was doing.
So if you want to face possible data corruption then use the option -9 without even thinking.
If you want to reduce / remove possible corruption of data, NEVER use -9, EXCEPT if there are no other options.





152
MP Server / Re: CMOD 9.5 on RHEL with remote Oracle
« on: July 14, 2016, 03:13:14 PM »

153
MP Server / Re: Mapping single folder field with multiple AG fields
« on: July 14, 2016, 03:10:06 PM »
This is possible as long as the fields are of the same type.
We have situations where folders have 15 or 20 application groups mapped to them and the Report ID field is mapped 15-20 times to the application group RPT_ID field.

Add the desired Application Groups, and Applications as necessary, to the Folder and then simply map each field accordingly.

Hello Steve,

the question was not to map X AG into a folder.
The question was to have for one folder field the possibility to search at the same time at multiple field in the same application group.
You can do that, but you are limited to 2 fields.

The example is the following for example:

you have a folder field called "customer numero", and you want to search at the application group field "old_cus_nr" and "new_cus_nr" and "newest_cus_nr" and "very_old_cus_nr".
CMOD cannot do that yet, it can only look for 2 of them on the same AG. And you need to select in the folder field definition the option "Mapping Type" and choose the value "Operator Or" instead of "Single".

That was the question asked by the poster...

154
To run arsload in daemon mode:

arsload -h <instance> -f -d <data directory> -c <index directory> -t <sleep_time> -u <arsload_user_id> -G <field> -A <field>

This command is defined in a shell script that runs at startup/boot time.  We've been running arsload in daemon mode for over 15 years with multiple CMOD instances (10+) on a single server.


The world is wonderful :-) one part is using it for more than 15 years, and another part of the world is not using it at all :-)
I just love it!!!!!!! :-)

155
MP Server / Re: CMOD 8.5 on RHEL 6.2 with remote Oracle 11g database
« on: July 14, 2016, 03:02:12 PM »
All,
If any of you completed setting up CMOD with remote ORACLE db server?
if so, 1)Can you share all the CMOD configuration parameters that need to be taken care, and what values are assigned to them.

2) Do I absolutely need table creation exit with for having Oracle on remote server? Any working sample of tablespace creation routine available? IBM Document is not enough for me.

Hello SV,

Well, local or remote Oracle DB instance for CMOD that's the same... CMOD will need to be able to read the tnsnames.ora.

The value that you need to put in the file ars.ini for the SRVR_INSTANCE entry is the entry in the tnsnames.ora:

in tnsnames.ora (for example)
Code: [Select]
MYORCLDB=
 (DESCRIPTION=
   (ADDRESS = (PROTOCOL = TCP) (HOST = host.example.com) (PORT = 1521)
   (CONNECT_DATA=(SID=ODDB))
)

and in the ars.ini

Code: [Select]
[@SRV@_CMODTST]
HOST=
PROTOCOL=2
PORT=0
SRVR_INSTANCE=MYORCLDB
SRVR_INSTANCE_OWNER=ondemand
SRVR_OD_CFG=/opt/ibm/ondemand/V9.0/config/ars.cfg
SRVR_DB_CFG=/opt/ibm/ondemand/V9.0/config/ars.dbfs
SRVR_SM_CFG=/opt/ibm/ondemand/V9.0/config/ars.cache

ars.dbfs must be empty

For the ars.cfg, then you need to have to fill the following entries:

Code: [Select]
...

#
# Database for OnDemand to use (Library Server Only)
#
ARS_DB_ENGINE=ORACLE


###########################################
# ORACLE Parameters (Library Server Only) #
###########################################
ARS_ORACLE_HOME=/oracle/product/11.2.0.2
...

Or you can read this document from IBM http://www-01.ibm.com/support/docview.wss?uid=swg27019582

Or simply read the official documentation of CMOD in the knowledge center http://www.ibm.com/support/knowledgecenter/SSEPCD_9.5.0/com.ibm.ondemand.installmp.doc/dodlx093.htm

You don't need to create the user tablespace/table exit. But that's an advise. Sample for tablespace are provided in the "exits" directory in the base installation directory of CMOD.
What do you need exactly for the tablespace user exit?

156
Other / Re: Ondemand Instance not starting
« on: July 14, 2016, 02:44:55 PM »
I don't see anything strange, except that:

Quote
I didn't make any changes to Trace.setttings file.
[TRACE]
COMPONENT_LEVEL=FFFFFFFFFFFFFFFFFFFFFF
#TRACE_FILE=ARCHIVE.trace.log
TRACE_FILE=ARCHIVE.trace.log
TRACE_FORMAT=TEXT
APPEND=0
MAX_LOG_SIZE=0

and

Quote
ARS_TMP=/sbcimp/dyn/data/RAR/OD_App1/temp
ARS_PRINT_PATH=/tmp
ARS_TRACE_SETTINGS=/opt/ondemand/config/checkod.txt
...
-rwxrwxrwx   1 odappmgr a19133         0 Jul  5 04:34 checkod.txt*

You said that you change the trace.settings, but the configuration (ars.cfg) shows that the trace file is called checkod.txt... that's not consistent.
either you have everything in checkod.txt, and that's ok...
or you change everything in trace.settings, and then you should change ars.cfg to be something like that:

Quote
ARS_TRACE_SETTINGS=/opt/ondemand/config/trace.settings

Otherwise that will never work!

157
Other / Re: Ondemand Instance not starting
« on: July 05, 2016, 12:07:06 AM »
Hello Gobi,

wow, it's back to the futur!! well back to the past...  8)

Ok... then first thing first.

What happened? Did it work before not starting anymore? What are the changes that happened that makes CMOD not working anymore? Something to do with an upgrade of your unix server?

Now to get more trace, you must modify your ars.cfg to be more like that:

Quote
ARS_TMP=/sbcimp/dyn/data/RAR/OD_App1/temp
ARS_PRINT_PATH=/tmp
ARS_TRACE_SETTINGS=/opt/ondemand/config/FILENAME

And the file called /opt/ondemand/config/FILENAME can be like that (exactly what you did):

Code: [Select]
[TRACE]
COMPONENT_LEVEL=FFFFFFFFFFFFFFFFFFFFFF
#TRACE_FILE=ARCHIVE.trace.log
TRACE_FILE=ARCHIVE.trace.log
TRACE_FORMAT=TEXT
APPEND=0
MAX_LOG_SIZE=0

One of the main error I get when CMOD cannot start are the following:

- Problem of permissions somewhere -> Check all the path that CMOD needs to access
- Problem with TSM, check that you don't have any error (again permissions) with TSM
- Problem with the DB, you need to ensure that your DB is not corrupted and working, and that everything is setup correctly, that the CMOD instance owner has dba rights on the database

What can help me from time to time, is the following, I change the file /opt/ondemand/bin/arslog to have something like that:

Code: [Select]
#!/bin/ksh

echo "$@" >> /tmp/myOD.log

exit 0

That way I can see if CMOD writes something in the System Log, which could help me to solve the problem.

Another thing that I do, even if it doesn't work always, is to use (Solaris/AIX) the "truss" command when running arssockd. In Linux you must use the command "strace". Then look for any problem with permissions, etc...

Hope that helps a little bit.

158
Well normally, ARSLOAD is never started at the beginning to load documents.
It means you MUST have a script to start it automatically at the start of the server, or have in the init.d (or this shç*t of systemd).

Normally to run it manually, you must run it with the option -d in order to look at which directory to look for files, and it should be also running in "background" waiting for files to be loaded, and also the -c option in order to tell arsload in which directory it must run the indexer.
Now what are exactly the whole parameter list you need, no idea. But it will be at least the argument -h and maybe -u/-p.

I've never used this functionality of arsload (in nearly 17 years and countless number of customer), so cannot help you more than that.
In case, you can look at the documentation of arsload for more help: (http://www.ibm.com/support/knowledgecenter/SSEPCD_8.5.0/com.ibm.ondemand.administeringmp.doc/arsload.htm) and maybe also http://www.ibm.com/support/knowledgecenter/SSEPCD_8.5.0/com.ibm.ondemand.administeringmp.doc/dodlo004.htm.

Good luck

159
MP Server / Re: Editable Index field/values in an AppGroup
« on: July 01, 2016, 04:58:04 AM »
Hello,

so first don't modify the table ARSAGPERMS :-) even if you could find out how it works, it is not supported by IBM to touch directly with some update sql commands.

That said, in order to be able to modify the index fields, you must ensure the following:
  • You have put the fields you want to update with the field type "updateable" (tab 'Field Information' in the AG)
  • The users have the permission to update the application group (tab "Permission" in the AG)

Then when both thing are done, then you can let the users change the fields.

Hope that helps.

160
Report Indexing / Re: ACIF OS/390 & USS
« on: June 15, 2016, 06:29:09 AM »
Hi Alessandro,

on Z/OS you have different ways to call arsload. I expect that also for your customer it is possible, to find a supported way to use arsload in Z/OS.

I'm looking with the IBM support :-) so I hope there is a nice way to solve it!

161
MP Server / Re: Storing dates earlier than 1st Jan 1970
« on: June 14, 2016, 09:35:54 AM »
just an example. You want to find a customer with name 'Schmidt' (a lot of results). In such a case it helps to know the birthdate  to find the customers documents.

It really depends on the data model that was used for defining the field in the AG. At the moment, EXCEPT for the example provided by Justin, which seems appropriate in this use case... the birthdate is never needed, even in the case of 10000 "Schmid", then you use the customer ID, which is unique.
And if you use the birthdate for that kind of filtering, then something is not correct in my own opinion.

But again, that's my opinion, and others have their own view on what should be used or not on a data model for an AG.
And this is also OK :-)

162
Report Indexing / Re: ACIF OS/390 & USS
« on: June 14, 2016, 09:28:10 AM »
just a question. Why do you want to run the arsload in USS and not as a job in Z/OS

Well I was asking that for one of my customer. apparently they are doing the stuff there in some unsupported ways, like running ARSLOAD from assembler.
Apparently the way they are doing now is using USS. And this has some limitation...

Well, I will look to open a enhancement request with my customer in order to solve that issue.

More than that, I am unable to answer, since I'm a noob in z/OS!! :D

163
Report Indexing / Re: ACIF OS/390 & USS
« on: June 14, 2016, 02:18:21 AM »
Sadly, that is still a restriction.

It's because the I/O that's used is based on the operating system that the OS/390 Indexer runs under.

Really sad ... :-(

164
MP Server / Re: Storing dates earlier than 1st Jan 1970
« on: June 14, 2016, 02:07:39 AM »
Mark today on your calendar -- I actually disagree with Alessandro.  :)

Some of my biggest customers are health care companies.  And one of them has made a huge effort to get everything that is a 'form' into CMOD.  So if you show up at an Emergency Room in a department, they ask you for your name and date of birth on almost every form.  It's the same way police identify individuals, because it's extraordinarily close to being unique.

So, it only makes sense that you search for someone by their name, maybe their "health card" number, and the 'verification' information is their birthdate.  It's also not outrageous to use the date field as a second criteria for common names like "Susan Smith" and "Michael Lee".

Now, being a consultant that has to deal with this transition, I totally wish it had been done a decade ago, before the problem was so much bigger, and so much more complex.

 ;D

Now that you give your example, I agree with you, such use case makes sense...
I have not worked with CMOD in the health industry, but mainly in the financial industry... and in that industry having the birthdate as a field makes no sense!!!

Thank you for your disagreement with me :-) I've learned something new today!!! :-) I know I will sleep like a baby tonight!!! ;D

165
iSeries / Re: Archiving Emails in OnDemand
« on: June 14, 2016, 01:54:41 AM »
Damn, that's sad.  CMOD is the perfect place for eMails -- uses data compression, consolidates multiple eMails into a single object, could store attachments as resources...

Am I missing something?

At least from IBM, there are no product anymore for that... maybe we can push it back through Enhancement Request... And with ODUG influence maybe it might come back... maybe.
If other products (read: Not IBM product) do that... I am willing to know it... because I have no knowledge of such precious item!

Pages: 1 ... 6 7 8 9 10 [11] 12 13 14 15 16 ... 65