DX Unified Infrastructure Management

We have seen a few situations where the Harvester upgrade from 9.0.161 to NFA 9.1.3 fails during the migrator stage of the upgrade with the same error.

The migrator is responsible for performing all of the mysql updates needed for each upgrade.

In this case the \Netqos\migrator\migrator.log file stated that it could not insert into poller.routers from harvester.routers and harvester.console_routers because the 'profileID' column did not allow a 'NULL' value.

The full error is:
FATAL TheSet::play() - Error playing back movie
com.ca.im.migrator.exceptions.MigratorException: Movie playback failed.
at com.ca.im.migrator.business.Projector.play(Unknown Source)
at com.ca.im.migrator.TheSet.play(Unknown Source)
at com.ca.im.migrator.TheSet.migrate(Unknown Source)
at com.ca.im.migrator.MigratorApp.main(Unknown Source)
Caused by: com.ca.im.migrator.exceptions.SqlActionExecutionException: SQL action failed
at com.ca.im.migrator.business.action.SqlAction.execute(Unknown Source)
... 4 more
Caused by: com.mysql.jdbc.exceptions.jdbc4.MySQLIntegrityConstraintViolationException: Column 'ProfileID' cannot be null

Solution
This can be corrected by changing the "left join" in the scirpt that loads up the poller database with just an "inner join". The offending script file can be found at '\<install_dir>\Netflow\sql\poller\toVersion9_1_0\populate-poller.sql' however you will need to update the '\NETQOS\collector_movie.xml', which needs to be edited to change the 'left join' to just 'join' as 'join' defaults to using inner join'.

To apply the chnages:

1) Go to the '\NETQOS\' directory and find the exisitng file named '\NETQOS\collector_movie.xml' and make a backup of the file.

2)Search for "left join harvester.console_routers" and remove 'left' from the file so that the full insert statement looks like below and save the file:
insert into poller.routers (address, state, statetime, profileid, discoverytime, routerpolltime, interfacespolltime, mappedtime, refreshtime, snmpversion, snmpport, snmpretries, snmptimeout, snmpmaxrepetitions, updatedon) select distinct cast(inet_ntoa(ho.Router) as char), IF(rr.firstpollerror=0,'Mapped','InitialPoll'), 0, rr.profileid, lastdiscovery,unix_timestamp(), unix_timestamp(), unix_timestamp(), lastrefresh, rr.snmpversion, rr.snmpport, rr.snmpretry, rr.snmptimeout, rr.snmpmaxrows, unix_timestamp() from harvester.routers ho join harvester.console_routers rr on ho.Router=rr.routerAddress

3) Open the Command Prompt and rerun the \<Install_Dir>\migrator\runMigrator.vbs script and check then check the \<Install_Dir>\migrator\migrator.log file to see if the migrator completes with exit code 0.

**Note you will only see this problem on upgrades from 9.0.161*

FYI this has been resolved in a newly release version of NFA 9.1.3 released on Support.ca.com on 11/12/2013.

If you are planning to upgrade to NFA 9.1.3 make sure you have the latest versions of the Harvester installation files data 11/12/2013.

Is there a new build number or is this repairing a botched release?

The new build number is 9.1.119.  You can get this from the Harvester databse in the revision_history table.

The ISO is DVD11170559M.iso and the old Harvester installs have been removed from the download site.

If you are on 9.1.118 already and didn't run into the error during an upgrade there should be no reason to go to 9.1.119, however if you have not upgraded yet from 9.0.161 you should make sure to download the current version that is on support.ca.com.