[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: c_get lo failure



--On Friday, July 20, 2007 5:36 AM -0700 Howard Chu <hyc@symas.com> wrote:

Howard Chu wrote:
Quanah Gibson-Mount wrote:
Anyone have any idea what would cause this error?

You have a corrupted index. No idea how that happened. Delete the index files and run slapindex...

=> key_change(ADD,5bf)
bdb_idl_insert_key: 5bf [0096defd]
=> bdb_idl_insert_key: c_get lo failed: DB_NOTFOUND: No matching
key/data  pair found (-30990)
<= key_change -30990
<= index_entry_add( 1471, "uid=btest,ou=people,dc=viz,dc=com" ) failure
bdb_add: index_entry_add failed
send_ldap_result: conn=6 op=3 p=3
send_ldap_result: err=80 matched="" text="index generation failed"
send_ldap_response: msgid=4 tag=105 err=80

At first, because of the "index generation failed" error, I thought it
was  related to file permissions, but the slapd user is completely able
to read  and write to files in the openldap DB directory.

Looking at this again, there's another possibility, and probably you should file an ITS for this. If you have a database with a zero-length suffix, the NULL context entry gets an entry ID of zero, which is usually an invalid ID number and always omitted from indexing. But it looks like with "slapadd -q" it's possible for this entry to get included in various indices, which will totally confuse them.

ITS#5055.

--Quanah

--
Quanah Gibson-Mount
Principal Software Engineer
Zimbra, Inc
--------------------
Zimbra ::  the leader in open source messaging and collaboration