Re: Stranges in databases, searches and mysterious dies...

[Turbo Fredriksson <turbo@nocrew.org>]

Quoting nazard@dragoninc.on.ca:

> On  9 May, Turbo Fredriksson wrote:
> > I've started to get strangest in my database. I can't search
> > for stuff that have been searchable before, and I can't bind to
> > the server...
> > 
> > And every now and then the slapd process are taking up a lot of CPU power,
> > and every now and then it also dies (during the night usually, so I don't
> > know WHY)...
> 
> Can't really help you with the CPU and crashing but...

I meant to say that SLAPD dies, not the CPU... :)

> > [papadoc.pts/3]$ ldapsearch -b 'uid=turbo,ou=People,dc=papadoc,dc=bayour,dc=com' objectclass=* sn -s sub
> > ldap_search: No such object
> > ldap_search: matched: OU=PEOPLE,DC=PAPADOC,DC=BAYOUR,DC=COM
> 
> These will probably fail because of filename expansion on your filter.
> Try using "(objectclass=*)"

Tried that,  but still the same...  And the shell  isn't expanding the
*... I  just tried the EXACT (copy  and past) same line  again, and it
worked...

> > [papadoc.pts/6]$ ldapsearch -b 'uid=turbo,ou=People,dc=papadoc,dc=bayour,dc=com' uid=turbo sn
> > ldap_search: No such object
> > ldap_search: matched: OU=PEOPLE,DC=PAPADOC,DC=BAYOUR,DC=COM
> 
> This will fail if your default scope is set to "one" (check ldap.conf).

It isn't, (unless it's the default, which I don't think it is).

> > ---- s n i p -----
> > [pts/4.papadoc]# ldapsearch -D 'uid=turbo,ou=users,dc=papadoc,dc=bayour,dc=com' -W uid=turbo sn
> > Enter LDAP Password: 
> > ldap_bind: Invalid credentials
> > ---- s n i p -----
> 
> Here you are using ou=users, not ou=people.

Typo in mail, not in command line, sorry...

> This should have worked (i think, still concerned about the objectclass
> thing)
>  
> > This is HIGHLY strange, since it worked before the reboot, and I have 
> > tried change the password with ldappasswd... Can still login with
> > the new password, but I can't bind to the LDAP database!
> > 
> > Before the crash I had 'dbcachenowsync', which I removed when the machine
> > was back online... The index have been rebuilt.
> 
> If you are concerned that the database may be corrupt, just rebuild it.
> (shutdown slapd, run ldbcat, mv database.dir database.old, mkdir
> database.dir, ldif2ldbm).

Did so  a couple of  times, didn't help.  I think the problem  (?) was
that the DB was to big. See the second update at the end of my mail...

> > index		uid,cn,sn,mail,mailalternateaddress,mailforwardingaddress,package approx,sub
> > index		uidnumber,gidnumber pres,eq
> 
> You might want an eq index on uid and cn.

I know. I have  that now, but I still get that  problem when/if I load
the Debian db...