zappa-sentry, automatic integration of zappa and sentry

Want to know when something goes wrong in a lambda? Tired of replicating all your alarm setup for each lambda?

If you just want a simple setup on each lambda that guarantees that you get alerts with enough information to be actionable, zappa-sentry is for you.

You’ll need a sentry project DSN. If you don’t have one or don’t even have an account you can create one for free at https://sentry.io/

How to use?

First, install the zappa-sentry in your lambda’s virtual environment with: pip install zappa-sentry (if you’re using a requirements.txt to manage dependencies don’t forget to add zappa-sentry to it).

Next, setup your sentry DSN as the value of environment variable SENTRY_DSN, either on the zappa_setting.json file or in any of the other methods on https://github.com/miserlou/zappa/#setting-environment-variables

Then you can setup the zappa_sentry.unhandled_exceptions handler.

Example:

{
    "dev": {
        ...
        "environment_variables": {
            "SENTRY_DSN": "https://*key*:*pass*@sentry.io/*project*",
            ...
        },
        "exception_handler": "zappa_sentry.unhandled_exceptions",
        ...
    },
    ...
}

And that’s all. Deploy your zappa function and you should see any errors appearing on sentry.

Want to get python/AWS tips on your email? Just subscribe!

Deleting old items in SQLAlchemy

Looking to delete old entries from a table because they’ve expired? Want to do it in an elegant way?

I usually like to split this kind of functionality into two different parts: the method that does the deleting and a static method that can be invoked from cron, a celery scheduled task or a django command.

As an example, let’s say we want to delete all the log entries on a system that are over 181 days (6 months) old.

Assuming a model like:

import datetime

class LogEntry(db.Model):
    __tablename__ = 'log_entries'

    id = db.Column(db.Integer, primary_key=True)
    text = db.Column(db.String(80))
    timestamp = db.Column(db.DateTime, default=datetime.datetime.utcnow)

First we add a method on the model that deletes expired log entries.

    @classmethod
    def delete_expired(cls):
        expiration_days = 181
        limit = datetime.datetime.now() - datetime.timedelta(days=expiration_days)
        cls.query.filter(cls.timestamp <= limit).delete()
        db.session.commit()

You’ll notice the use of @classmethod, that’s needed so we can invoke from the class and not from an object, as I’m doing in the next function (the one that can be called from a celery scheduled task, for instance):

def delete_expired_logs():
    LogEntry.delete_expired()

And with this you keep it elegant: all the relevant model relevant information in the model class, so if someone changes the timestamp field to another name, they will only have to change it in the delete_expired method, but you can easily call from somewhere else like a task or command.

Want to get python/AWS tips on your email? Just subscribe!