Horst Scheuermann
Horst.Scheuermann@uni-trier.de
Mon Jul 24 06:39:45 UTC 2006
> I cannot see why building dccd and dbclean with 64-bit ints and pointers > would by itself let the database become larger. it worked that way > The maximum size of the database is supposed to be controlled by > - a hard 3 GByte limit > If systems that can handle larger databases and larger mmap() > windows become common, that should be changed. > > - half of physical memory less 512 MByte, determined from > physmem_total(), sysconf(_SC_PHYS_PAGES), sysctl(mib, ...), > or pstat_getstatic(), depending on which is available > physical memory is 8GB > - the RLIMIT_AS, if available and sane > > - the RLIMIT_RSS, if available and sane not available > > My guess is that the last two are at issue. Does the `limit` or > `limits` shell command for the UID that runs dbclean and dccd say > for those two limits? > See the getrlimit man page ulimit -a core file size (blocks, -c) 0 data seg size (kbytes, -d) unlimited file size (blocks, -f) unlimited pending signals (-i) 65536 max locked memory (kbytes, -l) 32 max memory size (kbytes, -m) 6852955 open files (-n) 1024 pipe size (512 bytes, -p) 8 POSIX message queues (bytes, -q) 819200 stack size (kbytes, -s) 8192 cpu time (seconds, -t) unlimited max user processes (-u) 65536 virtual memory (kbytes, -v) 9276350 file locks (-x) unlimited -- mfG Horst Scheuermann 0651 201 3436 Universitätsrechenzentrum (URT) Abteilung zentrale Systeme 54286 Trier
More information about the DCC
mailing list