Bug#816294: [Pkg-openldap-devel] Bug#816294: slapd segfault on update dbconfig

2016-03-01 Thread Thomas Otto
Hello Ryan,


Am 29.02.2016 um 17:09 schrieb Ryan Tandy:
>>...
>> modifying entry "olcDatabase={3}hdb,cn=config"
>> ldap_modify: Other (e.g., implementation specific) error (80)
>>  additional info: failed to reopen database, rc=22
> 
> I performed the same modification on a new install of slapd 
> 2.4.40+dfsg-1+deb8u2, and did not observe a crash.

oh this is unfavorable for me ...

Our multimaster setup works since debian 7.x and this dbchanges also.
It occurs also in our test system with same setup, but no load.

I assume the problem is anywhere between the slapd and the new version
of bdb (v5.3).

I'll collect some config data for you.

> What was the olcDbConfig value prior to this change?

I only increase the value set_lk_max_locks from 5 to any other or
the same value with same effect.

> What sort of load was slapd under at the time? From the name, I guess 
> this might be an accesslog database?

yes but also occours on the main-db


I clean the DBs of one of my test systems. Now the Data synchronised again.

service slapd stop
rm -f /var/lib/ldap/*/*
service slapd start

... (wait some time) ...

ldapmodify ...

dn: olcDatabase={2}hdb,cn=config
changetype: modify
replace: olcDbConfig
olcDbConfig: set_cachesize 3 0 1
olcDbConfig: set_lk_max_locks   2
olcDbConfig: set_lk_max_objects 1
olcDbConfig: set_lk_max_lockers 1500

modifying entry "olcDatabase={2}hdb,cn=config"
ldap_modify: Other (e.g., implementation specific) error (80)
additional info: failed to reopen database, rc=22


I started slapd with debug output and attach this ...

I Attach also a filterd config ...


I hope this helps ...


best regards

Thomas Otto

-- 
Thomas Otto, Dipl.-Inf.
Friedrich-Schiller-Universität Jena
Rechenzentrum
Am Johannisfriedhof 2
D-07743 Jena
Tel.: 03641/9-40530
Fax.: 03641/9-40630
# slapd -u openldap -g openldap -d 127




56d5743e do_bind: dn () SASL mech GSSAPI
56d5743e ==> sasl_bind: dn="" mech= datalen=32
56d5743e SASL Canonicalize [conn=1003]: authcid="ldapadmin1/admin"
56d5743e slap_sasl_getdn: conn 1003 id=ldapadmin1/admin [len=13]
=> ldap_dn2bv(16)
<= ldap_dn2bv(uid=ldapadmin1/admin,cn=GSSAPI,cn=auth)=0 
56d5743e slap_sasl_getdn: u:id converted to 
uid=ldapadmin1/admin,cn=GSSAPI,cn=auth
56d5743e >>> dnNormalize: 

Bug#816294: [Pkg-openldap-devel] Bug#816294: slapd segfault on update dbconfig

2016-02-29 Thread Ryan Tandy

Control: tag -1 moreinfo

On Mon, Feb 29, 2016 at 03:57:30PM +0100, Thomas Otto wrote:

ldapmodify -Z -h ldaptest1
...
dn: olcDatabase={3}hdb,cn=config
changetype: modify
replace: olcDbConfig
olcDbConfig: # Note: most DB_CONFIG settings will take effect only upon 
rebuilding
olcDbConfig: # the DB environment.
olcDbConfig: #
olcDbConfig: # one 0.25 GB cache
olcDbConfig: #set_cachesize 0 268435456 1
olcDbConfig: set_cachesize 1 0 1
olcDbConfig: # Data Directory
olcDbConfig: #set_data_dir db
olcDbConfig: # Transaction Log settings
olcDbConfig: set_lg_regionmax 262144
olcDbConfig: set_lg_bsize 2097152
olcDbConfig: #set_lg_dir logs
olcDbConfig: # delete Transaction Logs:
olcDbConfig: set_flags DB_LOG_AUTOREMOVE
olcDbConfig: # Note that if the log files are removed automatically, recovery 
after
olcDbConfig: # a catastrophic failure is likely to be impossible.
olcDbConfig: # http://wiki.zimbra.com/wiki/OpenLDAP_Performance_Tuning_5.0
olcDbConfig: # db5.1_stat -c -h /var/lib/ldap/logs
olcDbConfig: set_lk_max_locks   6
olcDbConfig: set_lk_max_objects 25000
olcDbConfig: set_lk_max_lockers 1500

modifying entry "olcDatabase={3}hdb,cn=config"
ldap_modify: Other (e.g., implementation specific) error (80)
additional info: failed to reopen database, rc=22


I performed the same modification on a new install of slapd 
2.4.40+dfsg-1+deb8u2, and did not observe a crash.


What was the olcDbConfig value prior to this change?

What sort of load was slapd under at the time? From the name, I guess 
this might be an accesslog database?


I don't know for sure whether it's relevant, but for completeness: do 
you have other hdb databases, and what do their configs look like?


If you can think of any other information that might help reproduce the 
crash, please mention it.


thanks,
Ryan