[dev.icinga.com #1638] OCILIB not parsed error while binding values in object_ids_with_insert

[icinga-migration]

This issue has been migrated from Redmine: https://dev.icinga.com/issues/1638

Created by Tommi on 2011-06-12 10:37:14 +00:00

Assignee: Tommi
Status: Resolved (closed on 2011-08-02 09:15:58 +00:00)
Target Version: (none)
Last Update: 2014-12-08 14:35:50 +00:00 (in Redmine)

Icinga Version: 1.10.0
OS Version: any

---

there is a strange error while using ido2db_get_object_id with insert in upcomming idoutils V1.5.
Binding the first value will raise an OCILIB-Error claiming statement is not prepared. But preparing finished successfully and a getsql on the statement handle just before will print the right statement. I already spend days on this issue without a finding the reason. Maybe its a threading isue or some memory segments are overwritten- i have no glue.

Changesets

2011-06-12 10:23:44 +00:00 by Tommi db74189

add oracle trace option to config and implemented via stored procedure #1314
improved oci error handler
add more debug logging to catch prepared error #1638
refs #1314,#1638

2011-06-12 18:40:39 +00:00 by Tommi 934d920

workaround for object prepare error #1638
refs #1638

2011-07-13 17:19:38 +00:00 by Tommi 5ed6a8a

idoutils: fix increasing open cursors introduced with workaround for #1638
refs #1638

---

Relations:

• relates #1638

[icinga-migration]

Updated by Tommi on 2011-06-12 18:59:18 +00:00

• Category set to 61
• Status changed from New to Assigned
• Assigned to set to Tommi
• Done % changed from 0 to 50

workaround implemented(reparsing in every call).This is not a real solution.Reason is still unknown

[icinga-migration]

Updated by elippmann on 2011-07-11 15:48:55 +00:00

Hi there,

ocilib 3.9.0 is broken with "returning into" but it is fixed in ocilib 3.9.1.

Refer to http://orclib.sourceforge.net/public/changelogs/ocilib-changelog-v3.9.1.txt point 4.

Further, this workaround opens mass cursors since it does not free the extra prepared statements.

Regards

[icinga-migration]

Updated by Tommi on 2011-07-11 20:12:34 +00:00

Well, im checkin this, but we are using "returning" in several places and the "needs prepared" problem occures only with objects.

Expectation from oracle side with prepares are if we generate the same sql the hash with every reparsing in the same session will be equal and then only one cursor should be generated. But looks like its implemented in a different way.

Patch is already written, but needs to be tested before commit.

[icinga-migration]

Updated by Tommi on 2011-08-02 09:15:58 +00:00

• Status changed from Assigned to Resolved
• Done % changed from 50 to 100

for 1.5 i want to stay with the workaround expecting majoriti of users are not upgrading ocilib to safer version 3.9.2. Maybe it can be removed in a later release

[icinga-migration]

Updated by mfriedrich on 2014-12-08 14:35:50 +00:00

• Project changed from 18 to Core, Classic UI, IDOUtils
• Category changed from 61 to IDOUtils
• Icinga Version set to 1
• OS Version set to any