Hello, I am using ray.data to read some huge h5py file,which larger than 2gb, they are 8gb in fact.
And I treat the whole file as just one row of pyarrow table,which means one trunc, which may larger than the 2gb limit
But I am not sure for this,so I wish to dig deeper, then I found that I don’t know how to set log level of raylet,and they are info default,while I wish to set the log level to debug.
I start ray cmd is
ray --logging-level=debug start …
But this doesn’t work
And I found there are logs
[2023-08-08 10:55:25,824 I 1753438 1753486] (raylet) object_manager.cc:582: Received duplicate or cancelled chunk at index 25 of object 9953aed225098570ffffffffffffffffffffffff8400000002000000: overall 13601/108103 failed
[2023-08-08 10:55:25,829 I 1753438 1753485] (raylet) object_manager.cc:582: Received duplicate or cancelled chunk at index 682 of object 9953aed225098570ffffffffffffffffffffffff8400000002000000: overall 13602/108105 failed
But these logs are also appear for file smaller than 2gb,but not continuely ocurrs.