Any way to create fewer files after many runs?

[xerebz]

We are trying to use Devel::Cover to generate Perl coverage for many short lived perl file executions. The result is inodes being soaked up on the host system, on the order of 500k+ inodes. There are many files created in the runs subdirectory. Is there any way to compress the database after the runs so there are as few individual files as possible?

[xerebz]

Looks like it's caused by thousands of lock files. Can these be deleted after coverage generation but before coverage database merging and reporting?

/tmp/perl_coverage_db$ find . -name \*lock\* | wc -l
88289

[xerebz]

Looks like this is only part of the problem. There are also many individual files in the runs subdirectory. Can these be combined somehow?

find runs | wc -l
23364

[xerebz]

Asked another way, is there a way to add the coverage tallies to an existing DB? Instead of creating a new DB for every run?