Bug #16677

Journal size is too small when booting with the "debug" boot option

Added by segfault 2019-04-27 21:42:16 . Updated 2020-04-15 06:02:21 .

Status:
Confirmed
Priority:
Normal
Assignee:
Category:
Target version:
Start date:
Due date:
% Done:

0%

Feature Branch:
Type of work:
Code
Blueprint:

Starter:
Affected tool:
Welcome Screen
Deliverable for:

Description

After seeing Feature #12373, I expected to see debug messages from the greeter in the journal when booting Tails with the “debug” kernel command line option, but I don’t see any messages from the greeter in the journal, not even tails-greeter is ready. which should be logged with the INFO level (which should be the default level if I interpret data/tails-logging.conf correctly). Am I doing something wrong?


Files


Subtasks


Related issues

Blocks Tails - Feature #16925: Always add "debug nosplash" in troubleshooting mode Confirmed

History

#1 Updated by intrigeri 2019-04-28 08:26:10

  • Status changed from New to Confirmed
  • Affected tool set to Greeter

#2 Updated by intrigeri 2019-04-28 08:38:02

  • Assignee set to segfault
  • QA Check set to Info Needed

On Tails 3.13.1:

  • without passing debug, I see tails-greeter.desktop[3740]: [INFO] tails-greeter.py:127 usermanager_loaded: tails-greeter is ready. in the Journal.
  • when I pass debug, I see lots more debugging messages in the Journal

In both cases, I’m booting from an ISO in a VM, I set a admin password, and in a root terminal I run journalctl | grep -i greeter.

How can I reproduce your findings?

#3 Updated by segfault 2019-04-28 14:12:01

intrigeri wrote:
> On Tails 3.13.1:
>
> * without passing debug, I see tails-greeter.desktop[3740]: [INFO] tails-greeter.py:127 usermanager_loaded: tails-greeter is ready. in the Journal.
> * when I pass debug, I see lots more debugging messages in the Journal
>
> In both cases, I’m booting from an ISO in a VM, I set a admin password, and in a root terminal I run journalctl | grep -i greeter.
>
> How can I reproduce your findings?

I also can’t reproduce the exact result. But I still sometimes get fewer log lines than expected, see the screenshots. I’m wondering whether this could be due to rate limiting of journald.

#4 Updated by segfault 2019-04-28 14:12:12

  • Assignee deleted (segfault)
  • QA Check deleted (Info Needed)

#5 Updated by segfault 2019-04-28 14:13:05

  • Subject changed from Greeter doesn't write logs to Journal sometimes misses log lines

#6 Updated by segfault 2019-04-28 14:47:18

Now I noticed that on one machine which had the greeter lines, a few minutes later they were gone.

It seems like journald deleted the old logs because the maximum disk space was exceeded. The maximum disk space for journald is 10% of the filesystem that stores the logs. In Tails, that filesystem is /run. On my system, /run is mounted with a 200 MB tmpfs, so journald has 20 MB for logs. That doesn’t seem to be enough when booting with debug, because it’s used up after a short time.

#7 Updated by segfault 2019-04-28 15:02:27

We could consider storing the logs in /var/log instead of /run/log, by setting Storage=persistent or by creating the /var/log/journal directory, which will cause the default setting Storage=auto to use this directory instead of /run/log (see man journald.conf). In my VM, that would increase the maximum size for logs from 20 MB to 100 MB.

#8 Updated by intrigeri 2019-06-01 14:33:04

  • Subject changed from Journal sometimes misses log lines to Journal size is too small when booting with the "debug" boot option

#9 Updated by sajolida 2019-08-01 18:23:25

  • blocks Feature #16925: Always add "debug nosplash" in troubleshooting mode added

#10 Updated by sajolida 2019-08-01 18:26:43

Solving this ticket would make it possible to simplify a lot the troubleshooting instructions that we have in “Tails is not starting entirely”, see Feature #16925.

#11 Updated by intrigeri 2020-04-15 06:02:21

  • Affected tool changed from Greeter to Welcome Screen