Gorse worker refreshCache panic.

[binary4cat]

Please answer these questions before submitting your issue. Thanks!

Gorse version

Version:         v0.4.10-4-g762f21f
API version:     v0.2.7
Go version:      go1.19.4
Git commit:      762f21f3154f35877fe376ba435c11e7eaaf5edc
Built:           Fri Jan  6 09:00:06 UTC 2023
OS/Arch:         linux/amd64

Describe the bug
In the worker, the LastUpdateUserRecommendTime (worker/worker.go:1120) data in the cache is read in the refreshCache (worker/worker.go:1117) function, but for unknown reasons, it is not found, resulting in an error, but the error handling only determines the error except for the not found case, resulting in the timeLimit (worker/worker.go:1118) not being initialized to always be nil, which leads to a panic in the RemSortedByScore (worker/worker.go:1150) delete method calling timeLimit.

Since I don't understand the process here, I'm not sure if the not found error should be returned, so I need your help!

This is what I found after using mongodb, and it was not found when using redis and mysql.

To Reproduce
Steps to reproduce the behavior.

Expected behavior
A clear and concise description of what you expected to happen.

Additional context
Add any other context about the problem here.

[zhenghaoz]

Thanks for your feedback. What kinds of Mongo deployment do you use? Single node or cluster.

[binary4cat]

Thanks for your feedback. What kinds of Mongo deployment do you use? Single node or cluster.

Thanks for your work, I use a cluster with three nodes.

[zhenghaoz]

Thanks for your feedback. What kinds of Mongo deployment do you use? Single node or cluster.

Thanks for your work, I use a cluster with three nodes.

It is caused by inconsistent read between nods. I will give a better fix tomorrow.