Michael Schmitz wrote:
> I encountered a strange problem when building index files for the
current
> EMBL release: srsbuild coredumps when building the embl_cc.ids
file.
> At the time of the coredump, the file has the following size:
>> > 151484416 Nov 30 13:19 srs/index/embl_cc.ids
>> whereas the second-largest file is only 53 Meg in size.
>> The error message complains about insufficient memory; this may
explain the size
> of the partial index file to some extent.
the indexing has not changed a bit since 4.01. The "cc" index is
indeed by far the largest index of EMBL - one way to solve the
problem would be to simply disable the indexing of the comment
field. Another solution could be to improve the parser.
One last suggestion:
have you used
unlimit
before indexing? could be that the memory usage is simply
limited.
regards
thure
