Podman events with journald backend shows old events

[mheon]

Is this a BUG REPORT or FEATURE REQUEST? (leave only one on its own line)

/kind bug

Description

From the manpages of podman events:

If no since or until values are provided, only new events will be shown.

However, when I run podman events (F30, journald backend), I get old events.

Steps to reproduce the issue:

1. Unclear what prior steps are, but it is definitely necessary to create some events - start containers, remove them, etc. Ensure that the journald backend is set in the config file before you do this.

2. Run podman events with no arguments

Describe the results you received:

root@bellerophon code/libpod (fix_memory_leak) » podman events                                                                                                                                                                                                                  
2019-06-13 17:08:01.796349934 -0400 EDT container remove c0c861beca3d5f2e212ab6463dcb7eaa44e46f87b360e31a8987c549ba843439 (image=docker.io/library/fedora:latest, name=test1)
2019-06-18 16:20:43.211832187 -0400 EDT container start 171fdd7fde9f263da211db400c7791677b03e5dfc78819dbad0b311713512955 (image=docker.io/library/alpine:latest, name=confident_spence)
2019-06-19 17:19:26.057098614 -0400 EDT container remove 3c25d1d31b2b7d4b85c285cc1df2c290a55bc630e9f03ff3c08f976fd86d96e6 (image=docker.io/library/fedora:latest, name=agitated_sammet)
2019-07-05 16:27:33.612133295 -0400 EDT container remove 479534cb3a2ffd82c14441a72ee9b4968eda950cfbda142e3b39e5c0dca070a4 (image=docker.io/library/fedora:latest, name=test)
2019-07-16 09:25:44.348231845 -0400 EDT container remove a28a65dea7bbb2cb4bee0214b248447119fc89932a4c2dcb272e2efca5bbfe4b (image=docker.io/library/fedora:latest, name=thirsty_lewin)
2019-07-22 09:05:28.331977489 -0400 EDT system refresh

(Couple events from Friday, plus a refresh from my restarting the system this morning).

Describe the results you expected:

No events printed until new commands that generate events are run

Additional information you deem important (e.g. issue happens only occasionally):

Output of podman version:

Fedora packaged 1.4.4

[mheon]

It seems like it's always the last 6 events that occurred, so this is probably something on systemd's side?

[baude]

i doubt it; probably my mistake. I'll take a peek. One problem with the events though is we have to read the entire journal each time; I don't believe there is a way to get around that.

[rhatdan]

Does journalctl do this?

[baude]

it seems like the seek tail api is not honoring things. im looking into it but havent come up with a reason/solution yet.

[baude]

coreos/go-systemd#315

[rhatdan]

I am not sure how actively that package is maintained at this point.