Postgres Issue - Connection terminated unexpectedly | Railway | Page 1

fleet zealot Nov 7, 2022, 1:49 AM

#

Hello! I am currently using the postgres service and am having an issue where after a certain period of time my nodejs server returns "Connection terminated unexpectedly" when initially trying to hit postgres. Does anyone else have this issue or a solution for it? It works fine with a local db instance and on other services such as heroku

crisp goblet Nov 7, 2022, 6:57 PM

#

Hey @fleet zealot - may I have a Project ID?

#

How many connections would you say you have open?

fleet zealot Nov 7, 2022, 7:00 PM

#

Project ID:
b0c78ce2-647c-48be-89c1-7f4819e72bd9

#

and it should just be 1-2 open connections

crisp goblet Nov 7, 2022, 7:01 PM

#

Yea; thats strange.

#

I will see its a noisy neighbor.

#

Thats my guess.

fleet zealot Nov 7, 2022, 7:02 PM

#

ahh I see, that would make sense

#

yes, please keep me posted. Thank you very much!

crisp goblet Nov 7, 2022, 7:08 PM

#

Hmmm, how many times have you seen this?

fleet zealot Nov 7, 2022, 7:10 PM

#

It happens whenever I do not hit my api for an extended period of time. I had assumed it was sleeping in my staging env. But then it was happening in a prod env even with my account upgraded

#

once I hit the server once, it is fine and behaves as expected. But if I wait a few hours, it will happen again and give the same error

crisp goblet Nov 7, 2022, 7:12 PM

#

So looking at the logs I am getting

 
2022-11-03 01:35:19.738 UTC [665] LOG:  could not receive data from client: Connection reset by peer
2022-11-03 03:44:39.751 UTC [574] LOG:  could not receive data from client: Operation timed out
2022-11-03 03:46:50.823 UTC [673] LOG:  could not receive data from client: Operation timed out
2022-11-03 03:48:45.515 UTC [579] LOG:  could not receive data from client: Operation timed out
2022-11-03 03:48:45.515 UTC [576] LOG:  could not receive data from client: Operation timed out
2022-11-03 03:58:51.719 UTC [670] LOG:  could not receive data from client: Operation timed out

#

Your DB seems to be thrashing and I can't figure out why

#

Can you spin up a new DB in a different project and see if it still happens?

fleet zealot Nov 7, 2022, 7:22 PM

#

Yup! I will be home in a little while and can spin it up then and test. Thank you!

fleet zealot Nov 8, 2022, 2:19 AM

#

alright, I have spun up a new instance of postgres and my api and imported a db backup into it. so far so good, but I will test tomorrow morning and see if it has trouble connecting again after idle time. Will keep you posted 🙏

fleet zealot Nov 8, 2022, 12:50 PM

#

wow, same issue with a totally new project

#

project id if you can take a look at the logs:
02909059-bfb6-4461-a643-98a24dd0c06c

hidden aurora Nov 8, 2022, 3:54 PM

#

I am also having connection dropouts,
project id / deployment id :
3f69b4e7-32bb-4122-a64f-eabcb7e39ba2/fff131d

hidden aurora Nov 15, 2022, 5:45 PM

#

Has anyone else noticed improvements to this? Am still having issues myself and am planning the move to another DB provider

flint lotusBOT Nov 16, 2022, 5:04 PM

#

Flagging this thread. A team member will be with you shortly.

crisp goblet Nov 16, 2022, 6:05 PM

#

Hey everyone, I did an investigation with our team and I think we found an issue.

unfortunately, it’s due to some pesky neighbors on the infrastructure that should now be resolved. If you need SLAs for your databases, then I would urge you to upgrade to a teams plan.

hidden aurora Nov 17, 2022, 9:29 PM

#

Hey @crisp goblet , thanks for the help could you kindly confirm that using the 20 USD / seat team plan would eliminate these reliability issues?

Also could you kindly confirm if usage based costs are in addition to this per seat cost?

crisp goblet Nov 17, 2022, 9:36 PM

#

I can’t promise that there (things do go awry) but more so on the expectation of support and the use-case. If your DB is being used for a production env for a company, the Teams plan is the best way to ring our bell

#

It is 20 a month per seat plus usage, should be the same as your usage as it is now.

hidden aurora Nov 17, 2022, 9:39 PM

#

crisp goblet I can’t promise that there (things do go awry) but more so on the expectation of...

Right, I understand that you can't promise there won't be issues

But to the best of your teams knowledge of current issues you are tracking, do you reckon I am significantly much less likely to suffer disconnects on a teams plan? A best estimate / best effort answer is all I need and I obviously won't hold you accountable if there are issues

#

Or is this more so saying that I can get support quicker if there are issues?

crisp goblet Nov 17, 2022, 9:44 PM

#

hidden aurora Or is this more so saying that I can get support quicker if there are issues?

To this.

hidden aurora Nov 17, 2022, 9:45 PM

#

Thanks for the help and clarification

crisp goblet Nov 17, 2022, 9:46 PM

#

To give you a state of affairs; we have 110 open threads on this server, 20 or so in email. We are trying our best to make sure everyone is responded to in a timely manner. Upgrading to teams lets us know the difference between a hobby DB and a DB that is serving customers at scale. That way we can address your issues quickly.

#

Whatever the fix may be.

#

Its an anti pattern for us to be like: pay us or don’t have your thing work :’(

#

More so, help us prioritize how we direct our attention.

hidden aurora Nov 17, 2022, 9:48 PM

#

Understood yes, thanks again and yes that is of course understandable

crisp goblet Nov 17, 2022, 9:50 PM

#

Again, sorry about that, it personally feels terrible when I cant get to people on time. We are working on fundamental improvements here.

#Postgres Issue - Connection terminated unexpectedly