In the last few days, our largest index has started to fail during the
"writing word data" part of the process with an out of memory error. My
concern here is, of course, that we have hit some sort of upper level of
indexing capability. When this happens, the machine still has free
resources. So i guess I am wondering where to start the debug
process, or alternatively, if this is an easily solved solution, what
steps to take. Another index that we are building on the same set of
files, but with much tighter metaname/property organization, has not
encountered this issue yet.
Some notes about the setup.
OS: FreeBSD 5.4 STABLE, SMP, 2 x 2.3Ghz xeon, 2.5G ram
Swish-e version: 2.4.3 (will upgrade as a first step if this might help)
Indexing description: directory of xml files, total filecount in the
directory is 985707, looks like it started breaking at 984309. We aren't
using the btree/incrimental on this index, and are using -e, which had
helped us get past an out of memory issue that used to occur earlier in
the file parsing process.
i can post config files and such if that would be helpful, i am pretty
much looking for a starting point for the purposes of troubleshooting
As always, thanks for any assistance.
866 476 7862 x902
Received on Tue Oct 17 11:17:26 2006