Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: I built this Postgres logger

rocketgraph.io

Post reply on HN
Print view View original

41–49 of 49 posts

Re: Show HN: I built this Postgres logger

#41

Post by tetha »

koliber wrote: post #35
rsyring wrote: post #4

> In the future we can connect these logs to slack so you can get slack alerts when a developer accidentally DROPs a table. What can you do about getting me a slack notice BEFORE a developer accidentally drops a table? ;)

If you want to find the developer who will drop your table by accident, ask your devs to raise their hands if they ever dropped a table by accident. Those that don’t raise their hand are the ones you watch carefully.

Aye. A similar rite of passage: Any admin who hasn't nuked a server yet isn't working for long yet, or at their effective limit and is too careful.

We all have a few of these. My first one was a jenkins node with a "mv $WORKSPACE/ $WORKSPACE/bin" and no "set -u" :) Or accidentially wiping all SSH keys from a server due to a typo in the config management. That was fun.

But it's a good test for the mistake culture of the team and the resilience of their infrastructure. We had juniors nuke systems and no one but us and the monitoring noticed besides maybe a short hickup during a leader failover. That's pretty cool - and also a good horror story in a few words: "When you start ansible, and 3 minutes later on-call says 'Hi.' on slack".

Re: Show HN: I built this Postgres logger

#42

Post by palijer »

kvaranasi_ wrote: post #12
palijer wrote: post #10

Earlier quoted context omitted.

Haha, exactly, but if you have developers who can drop production tables without the org knowing in advance and reviewing the change, you already have a big problem. All our db operations like this go through our regular code review process for the db maintainers team, and instead of dropping tables, we just rename them with a prefix "to_be_dropped" as a scream test, then actually drop them a month later or so. Recov…

This is a nice idea to check if any dependencies are there on the table.

Haha, yeah, it's a hack we figured out the hard way, but it is super simple and effective.

Re: Show HN: I built this Postgres logger

#43

Post by gourabmi »

hobs wrote: post #39
kvaranasi_ wrote: post #29

Earlier quoted context omitted.

Of course I could be wrong. Please share with me your thoughts.

People using triggers for this just don't understand proper security - its a little harder but having your deployment process have the ability to drop tables and having your general user having read only query access (or insert powers on your own schema which nobody else can read) is stuff DBAs figured out in 90s and we keep forgetting.

What's a good place to start learning about these patterns ?

Re: Show HN: I built this Postgres logger

#44

Post by tracker1 »

koliber wrote: post #35
rsyring wrote: post #4

> In the future we can connect these logs to slack so you can get slack alerts when a developer accidentally DROPs a table. What can you do about getting me a slack notice BEFORE a developer accidentally drops a table? ;)

If you want to find the developer who will drop your table by accident, ask your devs to raise their hands if they ever dropped a table by accident. Those that don’t raise their hand are the ones you watch carefully.

Nope... but did check in the prod settings for the "local" config while working on a production bug close to the end of my contract. A couple weeks later another dev had dropped a production table as he'd used the local/dev connection string. Fortunately, every service request was logged and re-playable... so the DB was restored from backup, and every service request re-ran in order. Lost about a day of productivity for everyone using the app/service.

As to the config, this at least was pretty common for Web.*.config files in the .Net space in the mid 00s. I was never a fan of it, and even with the newer options in the space, prefer simpler environment variables these days.

Re: Show HN: I built this Postgres logger

#45

Post by tracker1 »

codegeek wrote: post #24
rsyring wrote: post #4

> In the future we can connect these logs to slack so you can get slack alerts when a developer accidentally DROPs a table. What can you do about getting me a slack notice BEFORE a developer accidentally drops a table? ;)

"BEFORE a developer accidentally drops a table?" Ideally, you don't want developer(s) to have access to destruct data especially things like DROP. Now, a lot of small businesses may not have the resource to have separate DBAs etc but the rule still applies. This is more a control problem than a code problem. Having said that, you could write triggers to stop DELETE or DROPs from happening if you cannot control the ac…

I've made it a personal rule for a while now, that I generally won't touch production once it is "online" ... I'll work initial setup, but after that, I'm happy to be on a screen sharing session walking someone else through what to do, so I don't have access and cannot touch it directly. Usually in favor of a CI/CD pipeline that will do regular releases. This at least lowers risk, but doesn't strictly prevent a DROP in a db/migration script.

All the same, I don't touch prod servers, and especially don't touch client prod servers.

Re: Show HN: I built this Postgres logger

#48

Post by kvaranasi_ »

bdcravens wrote: post #46

This is just a front-end to CloudWatch Log Insights, right?

Sort of. I created this page to just show what can be done with pgAudit. But what I do at Rocketgraph is more. I provide Authentication, Postgres and GraphQl console. Postgres logger is an extension for demo purposes. In the future I plan to add slack alerts when a developer DROP's a database accidentally. And much more. Configuring AWS RDS is a pain.

I wrote about it here: https://blog.rocketgraph.io/posts/install-pgaudit

That's why I automated the process. Now every project comes with Postgres and pgAudit configured, Authentication, GraphQL and front-end SDKs right out of the box so it will be easier to develop web applications.

Re: Show HN: I built this Postgres logger

#49

Post by hobs »

gourabmi wrote: post #43
hobs wrote: post #39

Earlier quoted context omitted.

People using triggers for this just don't understand proper security - its a little harder but having your deployment process have the ability to drop tables and having your general user having read only query access (or insert powers on your own schema which nobody else can read) is stuff DBAs figured out in 90s and we keep forgetting.

What's a good place to start learning about these patterns ?

Well, it depends on your database, but honestly the knowledge was handed down to me by database enthusiasts of the times before.

I was studying SQL Server at the time, so Kendra Little, Brent Ozar, that whole crew were a good starting point.

Once you learn the basics of the DCL for your database, a lot of the stuff just becomes a natural extension, and almost always you're just talking about the official documentation.

Usually you have a server and a database level identity, and manipulating those usually break out the connection vs object level permissions.

Once you understand the maps of those and grouping users, you can do stuff like detect a user did a bad thing and add them to a group of users who are default denied from the database, insert a row in a table, and roll back their changes. A logon trigger that does rude things to people's sessions, or certainty that your developers can view some parts of some data without exposing your secrets.

Post reply on HN