seek(:tail) does not work

[richm]

seek(:tail) does not work. It does not return the last entry in the journal. For example, run this script:

require 'systemd/journal'
require 'time'

now = Time.now
j = Systemd::Journal.new(path: '/var/log/journal')
j.seek(:tail)
if rc = j.move_next
    puts "how is it possible to move_next from the tail"
end
diff = now - j.current_entry.realtime_timestamp
puts "journal tail is #{diff} seconds or #{diff/86400} days behind now"
count = 1
while rc
    rc = j.move_next
    count = count + 1
end
diff = now - j.current_entry.realtime_timestamp
puts "we finally got to the tail after #{count} entries - time #{j.current_entry.realtime_timestamp} - last entry is #{diff} seconds behind now"

On my Fedora 24 system, I get the following output:

how is it possible to move_next from the tail
journal tail is 7957356.183605957 seconds or 92.09902990284672 days behind now
we finally got to the tail after 39 entries - time 2017-01-06 15:28:12 -0700 - last entry is 80.451257957 seconds behind now

That is, when doing seek(:tail), it is really 39 entries behind the actual tail.

Compare this to the behavior of journalctl -f which returns the last entry

[ledbettj]

Hi, thanks for opening the issue.

According to the upstream documentation, seek(:tail) positions the read pointer immediately after the last item in the journal. As such, you should call move_previous before reading any entries. Calling move_next will cause you to see random entries.

It seems that this is not mentioned in the gem documentation 😨 , so we can probably improve that here.

[ledbettj]

In my testing if you add move_previous immediately after your seek(:tail) in your test program you will see:

we finally got to the tail after 1 entries

Let me know if this is not the case for you.

[richm]

Works for me. Thanks!