MultiZarrToZarr and asycnio issues

[groutr]

When attempting to use MultiZarrToZarr (kerchunk v0.2.9 with python 3.11/3.14, fsspec 2025.9.0) to combine a set of json files into a single zarr store, I consistently get errors from asyncio that the loop is not running. Am I missing something, or is there something missing from the documentation? Is there a specific way I need invoke the translate() method? As far as I can tell, the kerchunk documentation doesn't mention asyncio anywhere.

This snippet is being run from a script, an ipython console, and a jupyter notebook. In all 3 cases I get the same error.

from kerchunk.combine import MultiZarrToZarr
FILES = [...]  #  list of compressed json index files on S3
x = MultiZarrToZarr(FILES, concat_dims=['time'])
x.translate()    # RuntimeError: Loop is not running

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
Cell In[3], line 1
----> 1 x.translate()

File ~\miniforge3\envs\py314\Lib\site-packages\kerchunk\combine.py:645, in MultiZarrToZarr.translate(self, filename, storage_options)
    639 """Perform all stages and return the resultant references dict
    640
    641 If filename and storage options are given, the output is written to this
    642 file using ujson and fsspec.
    643 """
    644 if 1 not in self.done:
--> 645     self.first_pass()
    646 if 2 not in self.done:
    647     self.store_coords()

File ~\miniforge3\envs\py314\Lib\site-packages\kerchunk\combine.py:381, in MultiZarrToZarr.first_pass(self)
    378 """Accumulate the set of concat coords values across all inputs"""
    380 coos = self.coos or {c: set() for c in self.coo_map}
--> 381 for i, fs in enumerate(self.fss):
    382     if self.preprocess:
    383         self.preprocess(fs.references)

File ~\miniforge3\envs\py314\Lib\site-packages\kerchunk\combine.py:282, in MultiZarrToZarr.fss(self)
    277 fs = fsspec.core.url_to_fs(
    278     self.path[0], asynchronous=True, **self.target_options
    279 )[0]
    280 try:
    281     # JSON path
--> 282     fo_list = fs.cat(self.path)
    283     fo_list = [ujson.loads(v) for v in fo_list.values()]
    284 except (IOError, TypeError, ValueError):
    285     # tries again sequentially in comprehension below

File ~\miniforge3\envs\py314\Lib\site-packages\fsspec\asyn.py:118, in sync_wrapper.<locals>.wrapper(*args, **kwargs)
    115 @functools.wraps(func)
    116 def wrapper(*args, **kwargs):
    117     self = obj or args[0]
--> 118     return sync(self.loop, func, *args, **kwargs)

File ~\miniforge3\envs\py314\Lib\site-packages\fsspec\asyn.py:76, in sync(loop, func, timeout, *args, **kwargs)
     73 # NB: if the loop is not running *yet*, it is OK to submit work
     74 # and we will wait for it
     75 if loop is None or loop.is_closed():
---> 76     raise RuntimeError("Loop is not running")
     77 try:
     78     loop0 = asyncio.events.get_running_loop()

RuntimeError: Loop is not running

[groutr]

I was able to get it working, but only after a significant amount of trial and error. I feel there has to be an easier way.

async def combine_zarrs_async(paths, refs):
    def run_multizarr():
        mzz = MultiZarrToZarr(
            paths,
            indicts = refs,
            concat_dims=["time"]
            )
        return mzz.translate()

    combined_ref = await asyncio.to_thread(run_multizarr)
    return combined_ref

async def zarr_mfdataset(files):
    refs = []
    for rj in fsspec.open_files(files, compression='zstd', protocol='s3'):
        with rj.open() as fin:
            refs.append(json.load(fin))

    combined = await combine_zarrs_async(files, refs)
    return combined