Timeout when using xarray to write data to zarr.storage.DBMStore multiple times which makes the whole zarr store unreadable

[QINQINKONG]

Hi. I'm trying to use Xarray to output files to zarr.storage.DBMStore. Since, the dataset is quite large, I calculate and output the results to one zarr.storage.DBMStore multiple times (append dim='time'). However, when the program failed because of timeout (I request 4hours from the queue every time), I seems to unable to access the data that have already been written to this zarr store during previous times. I wonder how to dealing with issues like this?
More generally, what we should do if we encounter some error when using xarray to write to a zarr store multiple times along a dimension?

Below is my code to write and read files to and from zarr store, and also the error message after a timeout attempt:

code of writing data to zarr store

import zarr
import xarray
store= zarr.DBMStore('/scratch/brown/kong97/WBGT/cmip6/GFDL_CM4_WBGT.bdb', open=bsddb3.btopen)
#doing some calculations and get the output: WBGT
WBGT.to_zarr(store,append_dim='time')
store.close()

After several times writing to this zarr store, the program fails because of timeout (the cluster may become very slow for some reason that I don't know). Then I found that I cannot read the data that have been written into this zarr store previously.

store= zarr.DBMStore('/scratch/brown/kong97/WBGT/cmip6/GFDL_CM4_WBGT.bdb', open=bsddb3.btopen)
test=xr.open_zarr(store)
raise error: DBPageNotFoundError: (-30985, 'BDB0075 DB_PAGE_NOTFOUND: Requested page not found')

Problem description

It seems that the data written previously would be still accessible if I use the default directory storage. However, my cluster storage quota has a limit on the total number of files and using the default directory storage would create many small files and easily exceed limit.

Appreciate any help.
Thanks!