I am looking for the same but using New Relic (https://elements.heroku.com/addons/newrelic)

"Ability to alert when going above threshold and the process is killed"
From @Nebula on Discord
We also might just wanna have the cap not kill those processes

Would log egress also support Logtail.com?

Igor Gassmann Yes in theory. Though would love to know what you like about Logtail. I've fiddle with it and it seemed like just a much-less-featureful-yet-prettier DataDog

Igor Gassmann Yes in theory. Though would love to know what you like about Logtail. I've fiddle with it and it seemed like just a much-less-featureful-yet-prettier DataDog

Jake Cooper: It's good as a starting point for small teams/startups. I've setup Datadog in the past, and it indeed has way more capabilities. However, developers often struggle using it and complain about how complex it is. Logtail is simpler to use for teams that are getting started with telemetrics/monitoring.

I want to move from Heroku, where Papertrail logging is an easy (and inexpensive) add-on. Will providers be able to create add-ons for Railway easily?

New Relic pls

+1 for logtail support!

https://www.axiom.co support could be looked at too

https://www.axiom.co support could be looked at too

Raimond Lume: CTO of Axiom here… would totally commit to making this a thing

https://www.axiom.co support could be looked at too

Seif Lotfy: Yoyo! Founder of Railway here
We're gonna roll out an API in October
It likely won't include the Logging Egress stuff just yet but, once we launch that would love to have you sit down with us and jam on it!

https://www.axiom.co support could be looked at too

Awesome, I'll reach out via an alternative channel :)

Another +1 for Logtail. We switched from DD because Logtail was cheaper and had all the features we needed.

I would love to start migrating our apps from Heroku to Railway, but Logtail support would be essential for some of our apps.

would love to see a log drain to Datadog

Adding an update here. It seems that Regions + Networking will take priority for now, in the short term, we will be working on guides to enable Log exfil for those who need a solution yesterday. Added an est. date and an owner. Keep the comments coming!

Adding an update here. It seems that Regions + Networking will take priority for now, in the short term, we will be working on guides to enable Log exfil for those who need a solution yesterday. Added an est. date and an owner. Keep the comments coming!

Angelo Saraceno: Thanks for the update. Definitely interested in having a solid log egress workaround (and preferably stable search across deployments).

Igor Gassmann Yes in theory. Though would love to know what you like about Logtail. I've fiddle with it and it seemed like just a much-less-featureful-yet-prettier DataDog

Igor Gassmann: I haven't dug deeper yet but it feels that Logtail is better suited for solo dev or small teams who need a lighter solution to monitor their project without all the complexity Datadog offers.

Adding an update here. It seems that Regions + Networking will take priority for now, in the short term, we will be working on guides to enable Log exfil for those who need a solution yesterday. Added an est. date and an owner. Keep the comments coming!

Angelo Saraceno: Definitely understand the regions priority (highly expected :)). Would also love an escape hatch while the egress comes to light so that we don't ship our product blindly in prod.

We've shipped a refined logging experience in the "Observability Panel"

Let us know what you think/what else you want to see!

Would love to be able to feed out all logs to my own syslog servers for ingestion by any tools I already use vs having to monitor railway site for apps here, and then other systems for other resources already captured there.

We've shipped a refined logging experience in the "Observability Panel"

Let us know what you think/what else you want to see!

Jake Cooper: Thats kind of sexy! Checking it out now

I'd also love to see New Relic as an option.

We've shipped a refined logging experience in the "Observability Panel"

Let us know what you think/what else you want to see!

Great addition to the on-platform logging, but any update to if/when Log Egress would be supported?
We are hoping to be able to integreate for example https://axiom.co/ to our services.

I am also wanting to move from Heroku, but lack of log drains is a complete non-starter.

We need to be able to set up alerts in our log output based on regex patterns, etc.

We already have all this functionality set up in Logtail/BetterStack, and would either want to continue using that tool, or a first-class tool within Railway if available.

It's a little mind-boggling how this issue has sat open for two years. Log drains seem to me to be a fundamental part of any modern PaaS.

It's a little mind-boggling how this issue has sat open for two years

big agree

+1 new relic

Having the ability to integrate with SigNoz/OpenTelemetry would be very cool.

I know this can be done on an application level, just a thought.

We've shipped a refined logging experience in the "Observability Panel"

Let us know what you think/what else you want to see!

This is a great start, but needs to be coupled with the ability to create dashboards and alerts based on the logs. This is currently what we use Google Cloud's built in logging, or BetterStack for. For instance, we run queries against the logs to see if there are any critical errors happening, then fire off pages or emails based on those outputs.

If log drains aren't in the picture, and Railway wants to own that entire experience, you need to have something on par with BetterStack's Uptime / Logging Dashboards + Alerts.

Hey just wanted to add in here

I love your product but logging is really fucking broken

Sometimes logs are cutoff / fail to show

this is destroying my observability

Please fix - I like it here but this is really killing me

Sometimes logs are cutoff / fail to show

This is often a case of buffering logs, make sure you aren't buffering any logs.

I don't think I'm buffering any logs

usually a redeploy fixes bad logs for a given instance

this is just killing my productivity - sometimes I have issues that can't be addressed until I run it locally and get a better idea of whats going on In the logs

I have never experienced this myself, I have even once ran an incremental counter that counted up and printed the numbers to the logs, then had a monitor that would connect to their API and log if any numbers where missed, ran for hours without skipping a beat.

Maybe something on my end, but this is something I've been seeing pretty consistently these past few weeks

What logger are you using?

Log4j

After a quick google, it does indeed look like Log4j will buffer logs by default.

^ I'd check that your logs are not buffered.

Buffered logs will cause delay because they only output when the buffer is flushed - it effectively means that log lines are "batched up" and then shipped in batches (i.e. output X lines every N interval), which matches the behaviour you're describing.

Cool I'll look into turning that off