Invalid memory address or nil pointer dereference

[nikhillimaje]

Hello, we recently saw the below error when the pods were spinning up causing them to enter crash loop. The underlying data store was Redis and was degraded, because of the large number of connections from our service.

2025-02-10 21:37:29.931	
panic: runtime error: invalid memory address or nil pointer dereference
2025-02-10 21:37:29.931	
created by github.com/redis/rueidis.(*pipe)._background in goroutine 1998960
2025-02-10 21:37:29.931	
	github.com/redis/[email protected]/pipe.go:330 +0x1c
2025-02-10 21:37:29.931	
github.com/redis/rueidis.(*pipe)._background.func1()
2025-02-10 21:37:29.931	
	github.com/redis/[email protected]/pipe.go:430 +0x240
2025-02-10 21:37:29.931	
github.com/redis/rueidis.(*pipe)._backgroundWrite(0xd09f1ea000)

2025-02-10 21:37:29.931	
	github.com/redis/[email protected]/internal/cmds/cmds.go:145
2025-02-10 21:37:29.931	
github.com/redis/rueidis/internal/cmds.(*Completed).Commands(...)

Any idea what would be the exact cause for this error? Thanks.

[rueian]

Hi @nikhillimaje, this is probably duplicated with #702

please try the latest version.

[rueian]

Connections won’t be closed for no reason. Someone needs to close them.

[nikhillimaje]

Let's consider the above scenario where our application crashed because of the nil pointer. In this case as well, the connections won't be closed? Note that we're not explicitly calling client.Close() in our codebase.

[rueian]

In that case, those connections will be forced closed by your operating system.

[nikhillimaje]

Got it. Thanks for explaining.

Is there a way to enforce timeouts on the read requests? We have connWriteTimeout which as per the documentation is applied for the read requests as well, but I think it doesn't work in our case.

[rueian]

Use context deadline or context timeout:
https://github.com/redis/rueidis?tab=readme-ov-file#context-cancellation