Fix cache race #4
Conversation
| sync.RWMutex | ||
| mx sync.Mutex |
There was a problem hiding this comment.
Figured exporting the Lock/Unlock functions here was unintentional and a bad idea for a consumer to use.
There was a problem hiding this comment.
ah yes, it was just programmatic convenience!
| @@ -101,8 +101,8 @@ func (r *Resolver) LookupTXT(ctx context.Context, domain string) ([]string, erro | |||
| } | |||
|
|
|||
| func (r *Resolver) getCachedIPAddr(domain string) ([]net.IPAddr, bool) { | |||
| r.RLock() | |||
| defer r.RUnlock() | |||
| r.mx.Lock() | |||
There was a problem hiding this comment.
Switching away from RW mutexes may slow things down here. Given that we're caching against what would be a network request waiting on a mutex doesn't feel too bad, and if this becomes an issue we can always come back and optimize it.
There was a problem hiding this comment.
let's add a follow up issue? It shouldn't be hard to do, just queue the entries for cleanup and do that with write lock.
| sync.RWMutex | ||
| mx sync.Mutex |
There was a problem hiding this comment.
ah yes, it was just programmatic convenience!
| @@ -101,8 +101,8 @@ func (r *Resolver) LookupTXT(ctx context.Context, domain string) ([]string, erro | |||
| } | |||
|
|
|||
| func (r *Resolver) getCachedIPAddr(domain string) ([]net.IPAddr, bool) { | |||
| r.RLock() | |||
| defer r.RUnlock() | |||
| r.mx.Lock() | |||
There was a problem hiding this comment.
let's add a follow up issue? It shouldn't be hard to do, just queue the entries for cleanup and do that with write lock.
No description provided.