Error varaible is shared between multiple part of a lock when acquiring lock.

[dntam00]

Hello @rueian,
When trying to understand some logic of lock implementation, I'm curious about this line of code:

rueidis/rueidislock/lock.go

Line 334 in fc56243

if err = acquire(err, keyname(m.prefix, name, i), g.csc[i], force); err == nil {

Is it intentional to share this err variable when acquiring a lock?
If I have key majority is 2 and the first lock acquiring loop is error ErrNotLocked, then the second lock acquiring is automatically failed with same error without sending command to Redis.
Thank you.

[rueian]

Hi @dntam00,

Yes, it is intentional. The client tries to take a lock by setting the corresponding keys in sequence. For example, if your key majority is 2 and you try to acquire a lock "my_lock", it will send the following requests sequentially:

SET rueidislock:0:my_lock NX PXAT ts
SET rueidislock:1:my_lock NX PXAT ts
SET rueidislock:2:my_lock NX PXAT ts

If any of them fail with ErrNotLocked means the lock is still held by others and no need to continue.

[dntam00]

Hi @rueian
Thank you for your answer 😊.