[dev.icinga.com #252] Invalid object_id for custom variables in IDO

[icinga-migration]

This issue has been migrated from Redmine: https://dev.icinga.com/issues/252

Created by cdoebler on 2010-01-12 13:34:36 +00:00

Assignee: mfriedrich
Status: Resolved (closed on 2010-02-10 18:09:17 +00:00)
Target Version: 1.0.1
Last Update: 2010-02-10 18:09:17 +00:00 (in Redmine)

---

There are assigned invalid object_ids to custom variables.
Tested w/ host and service using icinga 1.0-stable and MySQL for IDO.

Config:

define host{
        use                     linux-server
        host_name               localhost
        alias                   localhost
        address                 127.0.0.1
        _host_custom_var        host_cv_value1
        }

define service{
        use                             local-service
        host_name                       localhost
        service_description             PING
        check_command                   check_ping!100.0,20%!500.0,60%
        _service_custom_var             service_cv_value1
        }

DB-output:

mysql> select * from icinga_customvariables;
+-------------------+-------------+------------+-------------+-------------------+--------------------+--------------------+
| customvariable_id | instance_id | object_id  | config_type | has_been_modified | varname            | varvalue           |
+-------------------+-------------+------------+-------------+-------------------+--------------------+--------------------+
|                 4 |           1 | 2147483647 |           1 |                 0 | HOST_CUSTOM_VAR    | host_cv_value1     |
|                 5 |           1 | 2147483647 |           1 |                 0 | SERVICE_CUSTOM_VAR | service_cv_value1 |
+-------------------+-------------+------------+-------------+-------------------+--------------------+--------------------+
2 rows in set (0,00 sec)

Regards,

Christian

Changesets

2010-02-08 20:09:15 +00:00 by mfriedrich c9985f5

fix oracle rewrites for other rdbms, add several fixes/features

* add database name in connection msg to syslog
* fix comment commentdata/downtimedata copypaste failures
* fix object insert rewrite for mysql/pgsql
* fix customvariablestatus timestamp

fixes #252,#272

[icinga-migration]

Updated by mfriedrich on 2010-02-08 19:56:17 +00:00

• Category set to Event Broker
• Assigned to set to mfriedrich
• Target Version set to 1.0.1

hmmm i defined one of my own, but i cannot reproduce the error

(1 server (timeserver), 2 services (testkisten))

mysql

mysql> select * from icinga_customvariables;
+-------------------+-------------+-----------+-------------+-------------------+-------------+------------+
| customvariable_id | instance_id | object_id | config_type | has_been_modified | varname     | varvalue   |
+-------------------+-------------+-----------+-------------+-------------------+-------------+------------+
|                 1 |           1 |      3357 |           1 |                 0 | CUSTOM_TEST | timeserver | 
|                 2 |           1 |      3557 |           1 |                 0 | CUSTOM_TEST | testkisten | 
|                 3 |           1 |      3946 |           1 |                 0 | CUSTOM_TEST | testkisten | 
+-------------------+-------------+-----------+-------------+-------------------+-------------+------------+
3 rows in set (0.00 sec)

pgsql also does something wrong:

Feb 8 19:47:57 nagbaer ido2db: Error: database query failed for 'INSERT INTO icinga_customvariablestatus (instance_id, object_id, status_update_time, has_been_modified, varname, varvalue) VALUES (1, 13413683899112883390, FROM_UNIXTIME(1265654876), 0, 'CUSTOM_TEST', 'timeserver')' - 'ERROR: integer out of range '

this customvariables patch costs me a lot of time grrrrrrrrrrrr

o_id is always object_id, contact_id, host_id or service_id => unsigned long
the parameter handed over to the function is just o_id => int
asprintf handles that correctly, but typecasting int => void => unsigned long fails heavy.

afterall it's a coding issue which gcc will show.

will be fixed in my oracle rewrite then.

coming to the table itsself:

icinga=# select * from icinga_customvariables;
 customvariable_id | instance_id | object_id | config_type | has_been_modified |   varname   |  varvalue  
-------------------+-------------+-----------+-------------+-------------------+-------------+------------
                 1 |           1 |      7791 |           1 |                 0 | CUSTOM_TEST | timeserver
                 2 |           1 |      7991 |           1 |                 0 | CUSTOM_TEST | testkisten
                 3 |           1 |      8380 |           1 |                 0 | CUSTOM_TEST | testkisten
(3 rows)

so it looks fine.

coming to oracle:

(huh found a bug in prepared statement)

I've also added debugoutput to show what gets inserted:

[1265657937.182843] [001.2] [pid=16093] ndo2db_save_custom_variablestatus() instance_id=1, object_id=20433, ts=(SELECT unixts2date(1265657935) FROM DUAL), modified=0, varname=CUSTOM_TEST, varvalue=timeserver

output is also fine from oracle.

         6            1      21021            1                 0   CUSTOM_TEST     testkisten
         4            1      20433            1                 0   CUSTOM_TEST     timeserver
         5            1      20633            1                 0   CUSTOM_TEST     testkisten

So i assume this has been fixed recently, but was nice to test because i found some other bugs coming through my oracle rewrite.

I also wrote a generic output which database is being used - while debugging 3 rdbm to see in syslog which one is being used.

will be applied to git master when housekeeping, oracle rewrites and delete fixes are through.

[icinga-migration]

Updated by mfriedrich on 2010-02-10 18:09:17 +00:00

• Status changed from New to Resolved
• Done % changed from 0 to 100
• Estimated Hours set to 2

nothing more to test, everytzhing else will be applied to git master soon.