Why does the Events parameter until need to be a negative value if a duration is used?

[arctic-alpaca]

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

/kind bug

Description

When using the podman CLI or API, the until parameter for the events command/endpoint can be a timestamp or a duration according to the documentation. When a positive value like 10s is used for the duration, podman returns immediately, even thought it is used in streaming mode. This seems to be caused by this and this line, which subtracts the duration from the current time and uses it as until timestamp. This isn't documented or intuitive.

Steps to reproduce the issue:

1. ./podman events --since 3m --until 5s returns immediately

2. ./podman events --since 3m --until -5s takes 5 seconds to return

Describe the results you received:
See previous point.

Describe the results you expected:
Documentation to use a negative value in until or the possibility to use a positive value.

Have you tested with the latest version of Podman and have you checked the Podman Troubleshooting Guide? (https://github.com/containers/podman/blob/master/troubleshooting.md)

The linked code line is in the current master.

[rhatdan]

Interested in opening a PR to fix?

@cdoern PTAL

[arctic-alpaca]

I am not familiar with Go and don't feel comfortable opening a PR.

[cdoern]

I will look at this today, the way podman has handled --until has been flaky in most scenarios until recently.

[cdoern]

@arctic-alpaca the lines podman events --since 3m --until 5s and podman events --since 3m --until -5s produce the same output for me, what version are you running on and is there anything else you are doing that could affect the output?

[arctic-alpaca]

@cdoern By output, do you mean they both "wait" 5 seconds for new events and then return?
The version should be at least 3.1 but I am currently away from my computer and won't be able to check for a week, sorry.
I'll let you know as soon as I'm back.

[cdoern]

on my main branch (synced with master) and on the current version of podman, both of these commands sit for 5 seconds and then output the requested data @arctic-alpaca

[Luap99]

@cdoern Can you try with a higher value for until. It could be that podman takes that long to read all your events.

I see this:

$ time bin/podman events --since 0s --until 5s

real    0m1.945s
user    0m2.437s
sys     0m0.453s
$ time bin/podman events --since 0s --until -5s

real    0m5.549s
user    0m2.446s
sys     0m0.419s
$ time bin/podman events --since 0s --until 10s

real    0m1.939s
user    0m2.437s
sys     0m0.436s
$ time bin/podman events --since 0s --until -10s

real    0m10.463s
user    0m2.523s
sys     0m0.433s

[cdoern]

thanks @Luap99, that will help me figure this out. I will get on it today