On Tue, May 17, 2011 at 2:02 PM, Noah Slater <nsla...@apache.org> wrote: > > On 17 May 2011, at 18:21, Paul Davis wrote: > >> Can you reproduce it? Which test was running when it failed? If you >> can reproduce it, can you attach gdb to your couchdb process to get a >> stack trace of when it segfaults? > > Okay, now we have: > > Program received signal EXC_BAD_ACCESS, Could not access memory. > Reason: KERN_INVALID_ADDRESS at address: 0x00000000107c3bc0 > [Switching to process 99177 thread 0x2a03] > 0x00000000107c3bc0 in ?? () > (gdb) > Continuing. > > Program received signal EXC_BAD_ACCESS, Could not access memory. > Reason: KERN_INVALID_ADDRESS at address: 0x00000000107c3bc0 > 0x00000000107c3bc0 in ?? () > (gdb) bt > #0 0x00000000107c3bc0 in ?? () > Cannot access memory at address 0x107c3bc0 > #1 0x00007fff86abe5b1 in EVP_DigestInit_ex () > #2 0x00007fff86a8c89d in ssleay_rand_bytes () > #3 0x00007fff86a8bf2e in ssleay_rand_pseudo_bytes () > #4 0x000000001074307a in rand_bytes_1 () > #5 0x00000000100fdc07 in process_main () > #6 0x000000001006ff5e in sched_thread_func () > #7 0x000000001018e8db in thr_wrapper () > #8 0x00007fff8909f4f6 in _pthread_start () > #9 0x00007fff8909f3a9 in thread_start () > (gdb) > > >
That's down in OpenSSL in what appears to be a call to `crypto:rand_bytes(N)`. In other words, something goofy is going on cause that bit of code is pretty core to couchdb and the various tests, so if it were widespread we would've heard reports. Maybe your fuzzy logic board needs replaced?
