When you want to make sure only one process modifies a given resource at a time you need a lock.
When you have more than one pod running in production to your server in your Kubernetes, you can't lock only in memory, you need a distributed lock.
Implementation using redlock
import Redis from 'ioredis';
import Redlock from 'redlock';
const redis = new Redis(config.REDIS_HOST);
export const redlock = new Redlock([redis], {
retryCount: 15,
});
export const lock = async (key: string, duration: number) => {
try {
const lockAcquired = await redlock.acquire([key], duration);
return {
unlock: async () => {
try {
await lockAcquired.release();
} catch (err) {
}
},
};
} catch (err) {
return {
error: 'Unable to get lock',
unlock: () => {},
};
}
};
Usage
const { error, unlock } = await lock('lockKey');
// error trying to get
if (error) {
return { success: false, error: String(error) };
}
try {
// modify a shared resource
await unlock();
return result;
} catch (err) {
await unlock();
return { success: false, error: String(err) };
}
Use cases
You can use a distributed lock to update the balance of a ledger account.
You can use it to avoid two processes consuming the same API at the same time.
In Conclusion
Distributed locks are one of many possible approaches to handle concurrency to shared resources. You could use a conditional put. Use a queue to process just one event at a time.
Investigate and discover what is the best solution for your specific scenario.
Woovi is an innovative startup revolutionizing the payment landscape. With Woovi, shoppers can enjoy the freedom to pay however they prefer. Our cutting-edge platform provides instant payment solutions, empowering merchants to accept orders and enhance their customer experience seamlessly.
If you're interested in joining our team, we're hiring! Check out our job openings at Woovi Careers.
Photo by Mackenzie Marco on Unsplash
Top comments (0)