DX NetOps

Hi, 

 When i am loading ssdb backup to spectroserver, showing 179000 mode loaded but after some days i see model has been increased 800000 automatically and i see physical memory utilization is 98%. In this time period, I didn't discover new devices So, please tell me why is it being. And give the solution.... Plzzz

You can run the following MySQL query against the DDMdb database (Archive Manager) to find the models that were created/discovered in a time frame.

By the way, what is the version of your CA Spectrum?

1. Open a bash shell (bash -login)
2. Navigate to $SPECROOT/mysql/bin directory

    cd mysql/bin

3. Launch the MySQL command prompt

    ./mysql -uroot -proot ddmdb (for Windows)

    ./mysql --defaults-file=../my-spectrum.cnf -uroot -proot ddmdb (for Linux or Solaris)

4. Run the following MySQL query:

     mysql> select m.model_h, m.mtype_h, u.user_name, m.landscape_h, m.model_name, e.type, FROM_UNIXTIME(e.utime,'%Y %D %M %h:%i:%s') from model m, event e, user_def u where e.model_h=m.model_h and e.user_key=u.user_key and e.type=0x10201 and e.utime > UNIX_TIMESTAMP('2017-09-01 00:00:00') and e.utime < UNIX_TIMESTAMP('2017-09-06 00:00:00') order by m.model_name;

Could be this known issue:

Juniper routers chassis reconfiguration (reconfiguring) causing SpectroSERVER cpu problems and duplicate chassis module… 

Thanks,

Silvio

Hi Silvio, 

I am using spectrum  10.1.1

Have you had a chance to run either the MySQL query or the CLI from the KB document?

I have run that MySQL query..... 

This is showing the output 

Empty set (0.00 sec) 

Plz give the exact solutions..... 

What was the MySQL query you ran? Did you modify the date range?

mysql> select m.model_h, m.mtype_h, u.user_name, m.landscape_h, m.model_name, e.type, FROM_UNIXTIME(e.utime,'%Y %D %M %h:%i:%s') from model m, event e, user_def u where e.model_h=m.model_h and e.user_key=u.user_key and e.type=0x10201 and e.utime > UNIX_TIMESTAMP('2017-09-01 00:00:00') and e.utime < UNIX_TIMESTAMP('2017-09-06 00:00:00') order by m.model_name;

Date range 2017-01-01 to 2017-09-06

Hi, 

Silvio.... 

The Archive Manager's OOB historical data retention is 45 days only, hence 2017-01-01 could not be the beginning date in the query. Use the date you restored the SSdb as the beginning date and date you noticed a high number of models as end date in your MySQL query.

Ok

Now again run query 

I use time range 2017-10-03 to 2017-10-09

Query is running and taking time

Hi Silvio, 

Query has been run and data fached. 

Now..... 

What should I do??? 

Now... What should I do??? 

Hello Silvio

What are those models that were modeled/discovered?

Are they Juniper devices?

If you cannot share the result in this community thread, please open a support case at CA.

They are not juniper... Cisco only

Dear Silvio, 

I can share all information as you require. Give the solution only. 

Hi Silvio,

 this is the output screenshot. so, what is the next step what should i do......models is increasing automatically. is there also any option to stop/disable model increment like through script or command . please reply with solution.

Hi Silvio...... 

Why don't you open an issue with CA Support? They should be able to help you.

Ok i am agree with you,  i can.... But why we are using community. Dear,  other people will also know about this issue/problem. This will help to other people also... Any problem buddy.... If you don't have solution.. No problem 

Unfortunately we cannot provide a solution as we don't know what the issue is.  More analysis needs to be done.   The most common reason we have seen this is due to the bug with the Juniper code as Silvio mentioned originally.

If the models are increasing, the easiest way to see what they are is to use CLI to show all models, then later in the day run it again and then diff the output. 

From <SPECROOT>/vnmsh:

./connect

./show models > models_before.out

Then many hours later:

From <SPECROOT>/vnmsh:

./connect

./show models > models_after.out

Diff the files and see what models/modeltypes have increased.  You can open a case and provide the CLI output.

Cheers!

Jay

Thanks