[Bug] Error when Adding Documents to DeepLake Dataset - LockedException

Question

[Bug] Error when Adding Documents to DeepLake Dataset - LockedException

mike2k11 opened this issue 10 months ago · comments

Description

Issue Description:

I encountered an issue while trying to add documents to a DeepLake dataset in an Azure container. I am the only user of this dataset, but I'm randomly getting an error that prevents me from adding more documents. The error message I received is as follows:

Error Message
deeplake.util.exceptions.LockedException: This dataset cannot be open for writing as it is locked by another machine. Try loading the dataset with read_only=True.

Steps to Reproduce:

Create a new DeepLake dataset in an Azure container.
Attempt to add more documents to the dataset using the following command:

from langchain.vectorstores import DeepLake
db = DeepLake.from_documents(
pages,
dataset_path=f"azure://{storage_account_name}/{container_name}/default_v4",
embedding=embeddings,
creds=creds
)

Expected Behavior:

I expect to be able to add more documents to the dataset without encountering any errors since I'm the only user of the dataset.

Actual Behavior:

I'm randomly encountering the LockedException error mentioned above, which prevents me from adding documents to the dataset.

Additional Information:

I am the only user of this dataset, so there shouldn't be any locking conflicts with other machines.
The error suggests using read_only=True when loading the dataset, but this is not a feasible solution for my use case as I need to add new documents to the dataset.
Environment:

Langchain Version: : 0.0.266
Python Version: 3.10.11

Screenshots:

Use Cases

No response

Ackermann Yuriy · Answer 1 · Sun Oct 15 2023 14:53:02 GMT+0800 (China Standard Time)

Can confirm

Ivo Stranic · Answer 2 · Mon Oct 16 2023 23:27:22 GMT+0800 (China Standard Time)

Hi @herrjemand Thank you for raising the issue. Is there any chance that you opened the dataset in another container or machine? That's critical because the locks are machine based, and not username-based.

Ackermann Yuriy · Answer 3 · Tue Oct 17 2023 00:05:39 GMT+0800 (China Standard Time)

No. I tried opening db, got segfault from activeloop with db locked. Could not unlock it at all. Had to delete dataset

…

On Tue, 17 Oct 2023 at 4:27 AM, Ivo Stranic ***@***.***> wrote: Hi @herrjemand <https://github.com/herrjemand> Thank you for raising the issue. Is there any chance that you opened the dataset in another container or machine? That's critical because the locks are machine based, and not username-based. — Reply to this email directly, view it on GitHub <#2547 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMPOFA3HT7YJCLZLT7M65DX7VG6JAVCNFSM6AAAAAA3WWZ3MKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONRUG42DANJVGQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

Nathan Voxland · Answer 4 · Tue Oct 17 2023 02:55:02 GMT+0800 (China Standard Time)

Can you send the complete error message? Knowing which lock is stuck will help diagnose the problem.

Nathan Voxland · Answer 5 · Fri Oct 20 2023 00:01:47 GMT+0800 (China Standard Time)

Are you able to provide any more information, @herrjemand or @mike2k11 to help troubleshoot?

mikayelh · Answer 6 · Thu Nov 30 2023 07:42:33 GMT+0800 (China Standard Time)

hey @herrjemand,

I hope you are doing well. I am checking in to see if the issue you were experiencing is still ongoing. If not, that’s perfectly fine, but our team is always happy to assist you in fully utilizing Deep Lake. I will close this issue for now, but if the problem persists, please feel free to respond to this email or continue chatting with us.

By the way, we will be releasing the new version of our Managed Tensor Database shortly, which abstracts away issues like this. Stay tuned!

Best regards,
Mikayel