Tails

#1 Updated by intrigeri 2019-11-10 15:30:59

Hi @Cozain,

so, about the fs.file-max thing:

> amnesia systemd[1]: Failed to bump fs.file-max, ignoring: Invalid argument

> proposed fix:

> https://github.com/systemd/systemd/issues/12803
> https://github.com/systemd/systemd/pull/12808

What’s the problematic impact of this error?

What makes you think it’s related to GDM failing to start? I see this error even in cases when GDM starts just fine.

> amnesia systemd[1]: /lib/systemd/system/tails-gdm-failed-to-start.service:11: Ignoring unknown escape sequences: “MAX_LENGTH=254 ; PREFIX=”Error starting GDM with your graphics card: " ; SUFFIX=“. Please take note of this error and visit ”$“:https://tails.boum.org/gdm for troubleshooting.” ; MAX_VIDEO_CARD_LENGTH=$(($MAX_LENGTH - $(echo -n “$PREFIX$SUFFIX” | wc -c))) ; VIDEO_CARD=$(lspci -d::0300 -nn | sed -E “s,.* VGA compatible controller \[0300\]: *,,” | cut -c “1-$MAX_VIDEO_CARD_LENGTH”) ; /bin/plymouth display-message —text=“$PREFIX$VIDEO_CARD$SUFFIX” "

FYI, this one is tracked by Bug #17166 already.

#2 Updated by Cozain 2019-11-10 18:11:32

> What makes you think it’s related to GDM failing to start? I see this error even in cases when GDM starts just fine.

the console message suggests it is starting tails-gdm-failed-to-start.service

> FYI, this one is tracked by Bug #17166 already.

good

#4 Updated by intrigeri 2019-11-11 11:50:22

Hi,

>> What makes you think it’s related to GDM failing to start? I see this error even in cases when GDM starts just fine.

> the console message suggests it is starting tails-gdm-failed-to-start.service

I don’t think so: when systemd starts, it tries to bump fs.file-max. Then the log shows that it loads unit files and reports syntax issues. But at this point it’s not starting the tails-gdm-failed-to-start.service service yet.

So, the only problem here is that systemd fails to bump fs.file-max. At this point, it’s unclear to me what the impact of this error is, aside of noise in the logs. Unless someone explains why this is a problem in practice, I’m tempted to close this bug and let the fix flow into systemd stable and then Debian Buster (or not): I trust the upstream and Debian maintainers to pick up the fix if it’s truly important. What do you think?

#6 Updated by Cozain1 2019-11-20 18:52:36

@intrigeri

I would say try to fix this because it is clearly affecting graphics:
“Error starting GDM with your graphics card”

tails has lots of graphics related issues and fixing this would not harm even if it turns out it is of little effect but combined with other issues it could add up and counterwise fixing it my lead to elimanating other issues

leaving it hanging around is not a good idea IMO because it is real

the proposed fix in the original post look trivial to apply

if you like to involve upstreamers with this you can, however if you think it is faster to process this on tails side then thats better

in any case it good to let debian know regardless, this helps to figure out if it is a more general issue and help more people

please let me know what you decide to do with this once you do

#7 Updated by Cozain1 2019-11-21 03:08:55

you also mentioned this is being already tracked in https://redmine.tails.boum.org/code/issues/17166

is this ticket then a duplicate or you would like to keep this to discuss upstream involvement?

you can copy my notes there if you like but keep the issue open (dont close both as I explained earlier)

#8 Updated by intrigeri 2019-11-22 10:53:10

> you also mentioned this is being already tracked in https://redmine.tails.boum.org/code/issues/17166

That comment of mine was about the “Ignoring unknown escape sequences” warning message, which is a different, unrelated problem.

#9 Updated by intrigeri 2019-11-22 11:01:27

Hi,

> I would say try to fix this because it is clearly affecting graphics:
> “Error starting GDM with your graphics card”

Do I understand correctly that the graphical Tails interface does not start on this computer?
If so, please report this separately: I don’t see a causality relationship between whether systemd manages to bump fs.file-max, and whether GDM is going to start successfully or not (on my test systems, I see the Failed to bump fs.file-max error message even if GDM starts just fine).

> the proposed fix in the original post look trivial to apply

If the fix applies cleanly on top of Debian Buster’s systemd, then yes, it should be easy to cherry-pick it initially. But then we have to maintain a forked systemd package for the next 2 years, which would take time that we won’t be able to put anywhere else. That’s why I would like to first understand what exact user-facing problem this would solve. At this point I still have no idea :(

> in any case it good to let debian know regardless, this helps to figure out if it is a more general issue and help more people

I’m happy to do this once I know what practical problem this failure to bump fs.file-max causes.