Well, since I wasn't sure what else to do I ran the server under valgrind and got a fault after about an hour, but it doesn't seem to be in the same routine. Maybe it will be interesting in any case, though. ==5988== Invalid read of size 8 ==5988== at 0xFAD78F4: delta_bind_bat (bat_storage.c:169) ==5988== by 0xFAD7A6D: bind_col (bat_storage.c:186) ==5988== by 0xF9921B4: mvc_bind (sql.c:1607) ==5988== by 0xF992957: mvc_bind_wrap (sql.c:1683) ==5988== by 0x4C66559: runMALsequence (mal_interpreter.c:651) ==5988== by 0x4C6AFEF: DFLOWworker (mal_dataflow.c:362) ==5988== by 0x35CB207EE4: start_thread (in /usr/lib64/libpthread-2.18.so) ==5988== by 0x35CAAF4B8C: clone (in /usr/lib64/libc-2.18.so) ==5988== Address 0x18 is not stack'd, malloc'd or (recently) free'd I looked at the code but am not sure what's going on, so rather than me guessing I hope this is of some value. By the way, when I rebuilt the server I used --enable-debug --enable-assert --with-valgrind but did not use --disable-optimize, so maybe this is suspect. I'll rebuild without optimization and have another go. Tim