Promtail 2.2.0 crashes when pushing to multiple Loki instances: "concurrent map read and map write"

[mxsiegle]

Hey,

we are running Promtail 2.2.0 as Pushserver (as upstream for other Promtail instances) and are trying to send the logs to multiple Loki instances.
As soon as I configure a 2nd Loki instance in the Promtail clients: section, it fails to start and crashes immediately with the following error:

level=info ts=2021-04-01T08:34:36.786819377Z caller=pushtarget.go:78 msg="starting push server" job=pushserver
level=info ts=2021-04-01T08:34:36.787261628Z caller=server.go:229 http=[::]:3500 grpc=[::]:3600 msg="server listening on addresses"
level=info ts=2021-04-01T08:34:36.788144852Z caller=kubernetes.go:264 component=discovery discovery=kubernetes msg="Using pod service account via in-cluster config"
level=info ts=2021-04-01T08:34:36.789287805Z caller=server.go:229 http=[::]:3101 grpc=[::]:9095 msg="server listening on addresses"
level=info ts=2021-04-01T08:34:36.789499173Z caller=main.go:112 msg="Starting Promtail" version="(version=2.2.0, branch=HEAD, revision=a27c79913)"
[...truncated...]
ts=2021-04-01T08:34:42.041918277Z caller=log.go:124 level=info msg="Re-opening truncated file /var/log/pods/default_nginx-ingress-local-controller-78c48fc9f6-jh664_27fe488e-01dc-496b-a037-510c6ac74acd/nginx-ingress-controller/0.log ..."
ts=2021-04-01T08:34:42.042127803Z caller=log.go:124 level=info msg="Successfully reopened truncated /var/log/pods/default_nginx-ingress-local-controller-78c48fc9f6-jh664_27fe488e-01dc-496b-a037-510c6ac74acd/nginx-ingress-controller/0.log"
fatal error: concurrent map read and map write

goroutine 296 [running]:
runtime.throw(0x2c03e70, 0x21)
	/usr/local/go/src/runtime/panic.go:1116 +0x72 fp=0xc00008bbc0 sp=0xc00008bb90 pc=0x43ae72
runtime.mapaccess2_faststr(0x29d64e0, 0xc000a21590, 0x2bd2073, 0xd, 0x14c1c950e, 0x444a360)
	/usr/local/go/src/runtime/map_faststr.go:116 +0x4a5 fp=0xc00008bc30 sp=0xc00008bbc0 pc=0x4179e5
github.com/grafana/loki/pkg/promtail/client.(*client).getTenantID(...)
	/src/loki/pkg/promtail/client/client.go:375
github.com/grafana/loki/pkg/promtail/client.(*client).processEntry(0xc000b1d040, 0xc000a21590, 0xc0117f48916f2051, 0x14c1d844a, 0x444a360, 0xc001485680, 0x222, 0x0, 0x0, 0x0, ...)
	/src/loki/pkg/promtail/client/client.go:406 +0x8c fp=0xc00008bc80 sp=0xc00008bc30 pc=0xd29a0c
github.com/grafana/loki/pkg/promtail/client.(*client).run(0xc000b1d040)
	/src/loki/pkg/promtail/client/client.go:234 +0x285 fp=0xc00008bfd8 sp=0xc00008bc80 pc=0xd27545
runtime.goexit()
	/usr/local/go/src/runtime/asm_amd64.s:1374 +0x1 fp=0xc00008bfe0 sp=0xc00008bfd8 pc=0x473ae1
created by github.com/grafana/loki/pkg/promtail/client.New
	/src/loki/pkg/promtail/client/client.go:198 +0x4b2

The full log is over 1000 lines long, so I attached it as a file:
Promtail_crash_multiple_targets.txt

The Promtail config looks like this:

clients:
  - url: https://<loki1>/loki/api/v1/push
    basic_auth:
      username: <user>
      password: <pass>
  - url: https://<loki2>/loki/api/v1/push
    basic_auth:
      username: <user>
      password: <pass>

I configured it with the syntax described in the docs: https://grafana.com/docs/loki/latest/clients/promtail/configuration/#supported-contents-and-default-values-of-configyaml

It crashes 99% of the time immediately after startup, but sometimes it randomly works and pushes the logs to both Loki instances. I assume it's a synchronicity issue with that map object (would match the error message too).
Let me know if you need any other infos.

Thanks.

[mxsiegle]

Thank you very much for the quick reaction.
I can confirm that v2.2.1 fixes the issue.