Recurrent error with Atlas Serverless MongoDB connection - Error: "unable to write wire message to network: write tcp"

[Tomas_Gomez (Tomás Gomez)]

I have several micro services connecting with a Mongo Atlas database, using different dbNames and collections. I’ve changed the service from the free tier Mongo Atlas version to the paid Serverless version. At the exact moment I moved to the Serverless version, occasionally (I would say twice per hour), all the services get blocked from the database and can not realize any operation over the MongoDB, receiving this specific error:

Failed with error: connection(“SERVER_URI”) unable to write wire message to network: write tcp IP:PORT->IP:PORT: write: broken pipe

There are some operations that I need to repeat instantly in order to not to lose information, and so I have two questions:

• ¿Does anyone knows why this error occurs and how could avoid it? (I could not found any info about it)
• ¿Is there a way of re injecting a db command in order to re run that specific command and not lose the data?

Any answer would be appreciated, thanks!

[Felipe_Garcia (Felipe García)]

I’m getting the same issue

[Tarun_Gaur]

Hello @Tomas_Gomez and @Felipe_Garcia ,

Welcome to The MongoDB Community Forums!

To understand your use case better, I would like to know a few details such as:

• Which Official MongoDB Driver are you using?
• Driver version
• Have you enabled retryable read/write?
• Any specific operations that you see are being done when you get this error?

You can take a look at below thread as a similar error has been discussed there

Connection(<project-id>.mongodb.net:27017) unable to write wire message to network: write tcp <some-ip> -> <mongo atlas ip>: write: broken pipe

MongoDB Go Driver engineer here: It’s expected that the Atlas Serverless infrastructure will sometimes close in-use network sockets. We typically expect that the retryable reads and retryable writes behaviors of MongoDB drivers should allow any in-progress operations to be retried on another connection automatically. @Jan-Gerrit_Harms or @Ian_Sebastian, do you know if retryable reads and writes are enabled in your MongoDB driver configurations? Note that retryable reads/writes are enabled by d…

Lastly, I would recommend you go through below documentation to understand more about retryable read/write operations and how MongoDB can you help you build resilient applications that can withstand network outages and failover events.

• Build a Resilient Application with MongoDB Atlas — MongoDB Atlas
• Retryable Writes
• Retryable Reads

Regards,
Tarun

[Maarten_Baijs]

@Tarun_Gaur, we have the same problem. We are using MongoDB Atlas serverless and a Go service running on GCP Cloud Run using the latest Mongo driver (v1.11.2).

“Error while executing aggregation pipeline: connection(dev-xxx-lb.xxx.mongodb.net:27017[-3]) unable to write wire message to network: write tcp xxxxx:51945->xxxx:27017: write: broken pipe”

Furthermore, we haven’t enabled retries explicitly, but the latest Golang driver has this enabled by default. I think this behavior can happen with any type of query, but most of the time we see it happen with aggregation pipeline queries, which make up the bulk of the queries we do.

The thread you referenced does seem to discuss a similar problem, but without any solution, though.

[Felipe_Garcia (Felipe García)]

@Tarun_Gaur

• Which Official MongoDB Driver are you using?
  MongoDB Go Driver
• Driver version
  v1.10.1
• Have you enabled retryable read/write?
  yes I have
• Any specific operations that you see are being done when you get this error?
  We had this error in find and update operations

thanks

[Tarun_Gaur]

I would advise you to bring this up with the Atlas chat support team. They may be able to check if anything on the Atlas side could have possibly caused this broken pipe message. In saying so, if a chat support is raised, please provide them with the following:

1. Cluster link / name which experienced the issue
2. Time & date including timezone for when it occurred
3. Exact error message output
4. Driver language and version

Regards,
Tarun

[xiongyan_zhong]

We also encountered the same problem, but it was not resolved by retrying. You can check if the maximum connection idle time (SetMaxConnIdleTime) is set. If this value is not set, the connection pool will not actively release some connections that have been interrupted by the MongoDB server.