the default shm size on railway is 64mb and there is no way to increase this

@Brody Yes, I know. Hence this question. Will wait for someone from team to suggest a solution to this.

ive talked to the team about this myself, they don’t seem too interested in this issue, you may need to host the database elsewhere

Wow! Didn't expect that. Ok, then I guess I need to move my database off of Railway!

yeah for your workload you would need to, as this is definitely a blocker they don’t seem too interested in solving unfortunately

Got it. Thanks Brody!

no problem, sorry for the letdown here

Are these any team members who checks the conversations in discord? or is it all community managed?

This is a community forum, you would need to be on Pro in order to get support from the team

but having Pro wouldn't help here as other Pro users have been left in the dark on this issue as well

Not really sure if this answers is satisfying, but I will ask the team directly as well on Slack. Should be as simple as passing a variable to the docker run command.

Yes in theory it's that simple, but they very likely don't have a mechanism in place that would allow them to tack on an extra flag to a specific users container run command

the team has directly told me and another pro user they can't offer custom shm sizes, so unless you are actually enterprise then you may get the same answer I did: you would need to tune postgres to not use as much shared memory

hey all! Brody is correct, you can't pass a flag to your start command, and we hadn't been able to prioritize a fix for this until recently, but I'm happy to report that we do have an environment variable now that enables control over shm_size. the variable name is RAILWAY_SHM_SIZE_BYTES which takes a value of bytes (as the name implies). let us know if you have any problems

I shall test that with my database!

edit, works perfectly!

@Bhanu Teja P, flagging Melissa's note above which should unblock you by allowing you to control shm_size using the RAILWAY_SHM_SIZE_BYTES variable.

@christian where do I add this env variable? In my repo which connects to the database? Or should I add this in the env variables of the database?

@Brody where did you add this? And how much did you set it to?

please do not tag the team, you add it to your service variables

Hmm… ok. Can you please share where that is? Service variables of the database?

correct

You can follow these steps:

1. Click on your database on the canvas

2. In the service settings pane, click Variables from the top menu

3. Click New Variable

4. Enter VARIABLE_NAME = shm_size // VALUE = your custom shm size

5. Click Add

Thanks Christian. Is there a way to directly reach out to your team when needing customer support instead of posting it on a community managed forum. Previously, it used to be so simple to contact the team via discord. Now it looks like my ticket automatically got marked as resolved without the team even seeing it

^

Yeah, but you said don't take pro as they would say the same thing. If I would have listened to that, I would have migrated to another service

I can only provide the information I am aware of, I did not know they were going to add this variable

the statement is true regardless

Hmm… ok, thanks.

I set the value to 512000000 (assuming this means 512 MB)

Hoping this will solve the problem

please don't set it anywhere near that high, as a hobby user that would likely be far above the limit for shm sizes on the hobby plan when they implement the limits

set it to 128MB, or 134217728 in bytes

I tried 128 MB, but it's still not enough for my usecase

the error above states you only went 12mb out of memory

I have no idea. I tried 128 MB, but it still throws the same error. I set it to 512 MB, and no error

let me set it back to 128 MB again and retry it

we may set the default to 160mb so try that

160 MB didn't work

Trying 256 MB next

please provide the errors

Same error

Raw query failed. Code: `53100`. Message: `db error: ERROR: could not resize shared memory segment "/PostgreSQL.1645299550" to 12615680 bytes: No space left on device`

yeah same error, but the number would change

It didn't change

My guess is there are multiple of these queries running in parallel, and together combined is more than 160 MB

otherwise, it would not have worked for 512 MB too

this is starting to sound not like a hobby usecase if I'm being honest

what's a hobby usecase?

Setting it to 256 MB worked. I am going to leave it at this.

that's good to hear

Thanks for all the help @Brody

no problem

Now, I don't have to write a migration script anymore 😅

I'll preemptively say this, when / if you use your 5gb volume up on the database, you will need to upgrade to pro

Where do I check this?

I currently see this

click on the volume attached to postgres

not even close to where you should be looking haha

This?

yep, looks like you are soon going to hit the limit soon, once that happens your database becomes inaccessible

In Pro, do I get a higher size? How much higher?

pro users get 50gb

Nice! Will upgrade to Pro

once on pro don't feel bad if you need to set your shm size higher

As soon as I upgraded to Pro, I see this

have you moved the project over to the pro workspace?

yeah

what do I do?

please confirm that the project is now located in the pro workspace

It is

show me what's in the drop-down for the workspace selector on the left of this screenshot

interesting, can you quickly check your transactions for the card you used to upgrade, where you charged $20?

Checking

There was an unpaid invoice. Paid it. Now, it's back to normal.

Any idea why this would still show 5 GB?

you would need to grow it, it's not done automatically as growing the volume means downtime, so it's left to the users discretion, you can grow it when you want to in it's settings menu

Understood. Again, thanks for all the help!

no problem