Re: Problem with ipfire build system

[Michael Tremer <michael.tremer@ipfire.org>]

[-- Attachment #1: Type: text/plain, Size: 2688 bytes --]

Hello Adolf,

> On 8 Dec 2024, at 12:04, Adolf Belka <adolf.belka(a)ipfire.org> wrote:
> 
> Hi Michael,
> 
> I have experienced a problem yesterday and today with the IPFire build system. Not a major problem but it only seems to have started recently.
> 
> Today when I tried to run the build command with the standard IPFire-2.x git clone I got the following error message.
> 
> ./make.sh build
> 
> ERROR: Not enough temporary space available, need at least 8192MiB, but only have 491 [ FAIL ]
>     Check /home/ahb/sandbox/ipfire-2.x/log_x86_64/_build.preparation.log for errors if applicable [ FAIL ]

So, there is a check that will ensure that there is 8 GiB available for the build directory so that (ideally) the build does not abort because of low disk space.

This is however a little bit of a gamble because we don’t always know exactly how much (temporary) space the build environment might need and it is a little bit difficult if a build is being restarted (because we might have consumed a chunk of the 8 GiB already).

Could you send me the output of “df -h” and “du -csh build_x86_64” when this problem happens?

> I was able to overcome it by running the ./make.sh clean command on the openvpn-2.6-meetup-rebased branch git repo that I also have on my system.
> 
> After running the clean command the normal ipfire-2.x git repo build command is working fine and no longer gives the above message.

It sounds like the calculation works well when there is no build directory. That is the easy case.

> Yesterday it was the other way around. I tried to run the build command on the openvpn-2.6-meetup-rebased branch and got the same sort of message as above just the amount available was a bit different. Something like 670 instead of 491 today. Yesterday, when I ran the clean command on my normal ipfire-2.x git repo then the openvpn-2.6-meetup-rebased git repo had no problem with the build. Just doing a reboot on my desktop machine that has the git repo's on it made no difference.

Hmm, is it possible that the build environment has gone so large that 8 GiB is exceeded and my calculation goes into negative numbers? That might break things hard.

> I have not had this problem in the past. Although not a major problem, I have definitely run a build on one repo when the other still had its build results in place so something has changed. Maybe something in Arch Linux with one of the recent updates I ran.

It shouldn’t. This code was changed when we introduced the containerised build system earlier this year. So I am quite confident that I broke this myself.

Best,
-Michael

> Regards,
> 
> Adolf.
> 
>