luca@OpenLDAP.org wrote:

Full_Name: Luca Scamoni Version: RELENG OS: Linux URL: ftp://ftp.openldap.org/incoming/ Submission from: (NULL) (82.63.140.131)

Testing releng got this segfaults in a master-slave setup on both machines. Both traces show the same:

(gdb) bt full #0 0x006d7860 in strncmp () from /lib/tls/libc.so.6 No symbol table info available. #1 0x0019085e in bdb_rdn_cmp (v_e1=0x17d7270, v_e2=0x9837d58) at cache.c:301 rc = 5 #2 0x080fc1f5 in avl_find () No symbol table info available. #3 0x00190abd in hdb_cache_find_ndn (op=0x17d7460, txn=0x9837aa8, ndn=0x17d73ec, res=0x17d7318) at cache.c:433 bdb = (struct bdb_info *) 0x97bf5d8 ei = {bei_parent = 0x97bf624, bei_id = 0, bei_lockpad = 0, bei_state = 0, bei_finders = 0, bei_nrdn = {bv_len = 23, bv_val = 0xffffffe9<Address 0xffffffe9 out of bounds>}, bei_rdn = {bv_len = 0, bv_val = 0x0}, bei_modrdns = 0, bei_ckids = 0, bei_dkids = 0, bei_e = 0x0, bei_kids = 0xb7f43994, bei_kids_mutex = {__m_reserved = 0, __m_count = 0, __m_owner = 0x0, __m_kind = 0, __m_lock = {__status = 0, __spinlock = 135208867}}, bei_lrunext = 0x55313c, bei_lruprev = 0x542022fc} eip = (EntryInfo *) 0x97bf624 ei2 = (EntryInfo *) 0x0 rc = 0 ptr = Variable "ptr" is not available.

The core happened on master shortly after starting the slave, then on slave shortly after staring the master again. Using HDB and syncrepl (two databases per instance, different suffixes)

I have access to the core(s) for any further information needed

Do you still have the data that caused this crash? Will it continue to re-occur each time you restart a down server? (Can you continue to reproduce the event...)

Please print *op and *eip as well. Also, what's in frame 4 or higher, what function called hdb_cache_find_ndn(), does the stack trace stop there or is there more?