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

Re: back-bdb quirk?

Jonghyuk Choi wrote:
> Pierangelo, increasing BDB_IDL_DB_SIZE worked in your case ?
> buf[BDB_IDL_DB_SIZE] of bdb_idl_insert_key can overflow while running
> bdb_idl_fetch_key().
> Although I'm also not 100% sure whether increasing BDB_IDL_DB_SIZE directly
> solves the problem,
> we seem to need a better management of IDL in memory.

I managed to repeat the test with current HEAD code and there was 
no problem now with slapadd; ldapadd with the same ldif file fails

conn=0 op=15268 ADD dn="cn=15266,dc=example,dc=com"
conn=0 op=15267 RESULT tag=105 err=0 text=
ber_flush: 15 bytes to sd 14
conn=0 op=15268 RESULT tag=105 err=0 text=
conn=0 op=15269 ADD dn="cn=15267,dc=example,dc=com"
ber_flush: 15 bytes to sd 14
conn=0 op=15269 RESULT tag=105 err=0 text=
conn=0 op=15270 ADD dn="cn=15268,dc=example,dc=com"
connection_operation: error: SASL bind in progress (tag=136643424).
Segmentation fault

I'll need to investigate this now; however I did it on a slightly 
different system (a RedHat 6.2 with a number of updates and fixes, 
so I don't know how reliable it is) than the one that failed (which 
was a RH 7.1 with standard updates). I won't be able to access it 
before saturday and I expect to be able to test it also on a 7.2.


Dr. Pierangelo Masarati               | voice: +39 02 2399 8309
Dip. Ing. Aerospaziale                | fax:   +39 02 2399 8334
Politecnico di Milano                 |
via La Masa 34, 20156 Milano, Italy   |