hi there.
I understand that I can use Loki and Grafana together to read logs from journald on a linux machine that uses systemd.
However, it isn’t obvious to me how I can see if logs for a given service were written to STDOUT, or to STDERR from a query in Loki.
I can’t seem to see any settings in the docs for promtail either:
Is this information exposed at all?
I know I can use promtail to read separate files, and use as a way to differentiate between to file streams, but if possible, I’d like to avoid doing this, because the journal in systemd has a number of other benefits too.
Oh, sorry, I think you’re right, @jangaraj. I am more familiar with Loki than Systemd, and I was looking at moving from Supervisor to Systemd for some services because I think Systemd offers a better fit in terms of
I had just assumed that this had been built into Systemd already, but it seems to be unevenly supported at best after looking around online.
I can do this in a running VM:
sudo journalctl _TRANSPORT=stdout
That gives me this output:
-- Logs begin at Thu 2024-02-29 15:46:46 CET, end at Tue 2024-05-07 18:39:01 CEST. --
Feb 29 15:57:21 app1.myapp.org 50-motd-news[3450626]: * Strictly confined Kubernetes makes edge and IoT secure. Learn how MicroK8s
Feb 29 15:57:21 app1.myapp.org 50-motd-news[3450626]: just raised the bar for easy, resilient and secure K8s cluster deployment.
Feb 29 15:57:21 app1.myapp.org 50-motd-news[3450626]: https://ubuntu.com/engage/secure-kubernetes-at-the-edge
(lots more lines)
I’m using sudo
here because the user I’m logged in as in thie case isn’t part of the adm
group that I think would give access to the logs.
When I set the transport to stderr
however:
sudo journalctl _TRANSPORT=stderr
I get this:
-- Logs begin at Thu 2024-02-29 15:46:46 CET, end at Tue 2024-05-07 18:39:08 CEST. --
-- No entries --
So there might be some kind of support, but it doesn’t seem to be official. I think it’s best for me assume that if I want the distinction between log output that STDOUT and STDERR would give me, I should use another approach, like careful use of log levels or so on.
Thanks for the quick respons BTW 
1 Like