Re: slapd memory fault

[matthew sporleder <msporleder%gmail.com@localhost>]

On 2/17/09, Sarton O'Brien <bsd-xen%roguewrt.org@localhost> wrote:
> matthew sporleder wrote:
>
> > On 2/17/09, Sarton O'Brien <bsd-xen%roguewrt.org@localhost> wrote:
> >
> > > ulimit -s 4096
> > >
> > >  NetBSD spike.internal 5.99.7 NetBSD 5.99.7 (XEN3_DOMU) #0: Sun Feb 15
> > > 00:54:47 EST 2009
> > >
> root%spike.internal@localhost:/usr/obj/sys/arch/amd64/compile/XEN3_DOMU
> > > amd64
> > >
> > >  6887      1 slapd    RET   write 1024/0x400
> > >  6887      1 slapd    CALL  lseek(8,0,0,0)
> > >  6887      1 slapd    RET   lseek 0
> > >  6887      1 slapd    CALL  fcntl(8,8,0x7f7fffffcf40)
> > >  6887      1 slapd    RET   fcntl 0
> > >  6887      1 slapd    CALL  write(2,0x7f7fffffbfb0,0x55)
> > >  6887      1 slapd    GIO   fd 2 wrote 85 bytes
> > >       "bdb_db_open: database \"dc=internal\": unclean shutdown detected;
> > > atte\
> > >        mpting recovery.\n"
> > >  6887      1 slapd    RET   write 85/0x55
> > >  6887      1 slapd    CALL
> > > __gettimeofday50(0x7f7fffffbec0,0)
> > >  6887      1 slapd    RET   __gettimeofday50 0
> > >  6887      1 slapd    CALL  getpid
> > >  6887      1 slapd    RET   getpid 6887/0x1ae7, 18220/0x472c
> > >  6887      1 slapd    CALL  fcntl(3,3,0)
> > >  6887      1 slapd    RET   fcntl 2
> > >  6887      1 slapd    CALL
> > > sendto(3,0x7f7fffffbee0,0x9b,0,0,0)
> > >  6887      1 slapd    MISC  msghdr: 48,
> > > 00000000000000000000000000a0ffff108c412
> > >
> 500a0ffff0100000000a0ffff0000000000000000a858b02400000000
> > >  6887      1 slapd    GIO   fd 3 wrote 155 bytes
> > >       "<167>1 2009-02-18T10:21:57.649703+11:00
> > > spike.internal slapd 6887 - - \
> > >        bdb_db_open: database \"dc=internal\": unclean shutdown detected;
> > > atte\
> > >        mpting recovery.\n"
> > >  6887      1 slapd    RET   sendto 155/0x9b
> > >  6887      1 slapd    CALL
> > > __stat50(0x7f7ffb0d1a60,0x7f7fffffd5a0)
> > >  6887      1 slapd    NAMI
> > > "/var/openldap/openldap-data/DB_CONFIG"
> > >  6887      1 slapd    RET   __stat50 0
> > >  6887      1 slapd    CALL  getpid
> > >  6887      1 slapd    RET   getpid 6887/0x1ae7, 18220/0x472c
> > >  6887      1 slapd    PSIG  SIGSEGV SIG_DFL: code=SEGV_MAPERR,
> addr=0x30,
> > > trap=6)
> > >
> > >  Yeah, I have no idea :)
> > >
> > >
> >
> >
> > Hmm.  It's not clear to me why your getting a SIGSEGV, but I can tell
> > you that bdb is trying to open DB_CONFIG and then re-run the __db.xxx
> > files to get the database back into sync.  Maybe try a db_recover or
> > even running slapd with debugging to see if it has anything else to
> > say (slapd -d -1?)
> >
>
>  Double hmmm. The only file other than DB_CONFIG was alock, due to a
> previous run.
>
>  A clean invocation with the options above:
>
>   5996      1 slapd    RET   write 1024/0x400
>   5996      1 slapd    CALL  lseek(8,0,0,0)
>   5996      1 slapd    RET   lseek 0
>   5996      1 slapd    CALL  fcntl(8,8,0x7f7fffffcf40)
>   5996      1 slapd    RET   fcntl 0
>   5996      1 slapd    CALL
> __stat50(0x7f7ffb0d1a60,0x7f7fffffd5a0)
>   5996      1 slapd    NAMI
> "/var/openldap/openldap-data/DB_CONFIG"
>   5996      1 slapd    RET   __stat50 0
>   5996      1 slapd    CALL
> __stat50(0x7f7fffffd100,0x7f7fffffd500)
>   5996      1 slapd    NAMI
> "/var/openldap/openldap-data/__db.001"
>   5996      1 slapd    RET   __stat50 -1 errno 2 No such file or directory
>   5996      1 slapd    CALL  getpid
>   5996      1 slapd    RET   getpid 5996/0x176c, 20291/0x4f43
>   5996      1 slapd    PSIG  SIGSEGV SIG_DFL: code=SEGV_MAPERR, addr=0x30,
> trap=
>  6)
>
>  Now it's been a while but I am pretty sure it should populate the
> directory?
>

You should have had those __db.xxxx files to keep your slapd database
functional and in-sync with dn2id and id2entry.  That's why I
suggested a db_recover.  See if that segfaults.