Release hdf5 disk memory after table or node deletion using pytables or pandas

Question

I am using HDFStore with pandas / pytables.

After deleting the table or object, the hdf5 file size remains unchanged. This space seems to be reused when additional objects are added for storage, but this can be a problem if large space is wasted.

I did not find any command in pandas and pytables APIs that can be used to recover hdf5 memory.

Do you know any mechanism to improve data management in hdf5 files?

Answer 1

see here

you need ptrepack it, which overwrites the file.

 ptrepack --chunkshape=auto --propindexes --complevel=9 --complib=blosc in.h5 out.h5 

as an example (this also compresses the file).