Initial installation - stuck at installing python (killing process)

Initial installation - stuck at installing python (killing process)

[dptdescribe]

[-- Attachment #1: Type: text/plain, Size: 1860 bytes --]

Hello there everyone,
For the last few days I am trying to install GuixSD on the virtual environment. I have already read few articles in this topic, official manual and even asked for help on the #guix channel. No luck so far.

Here is the process step-by-step (after preparing the VirtualMachine and adding):
ifconfig enp0s3 up && dhclientenp0s3
cfdisk /dev/sda
mkfs.ext4 -L root /dev/sda2
mount -L root /mnt
mkdir /mnt/etc
cp /etc/configuration/desktop.scm /mnt/etc/config.scm
nano /mnt/etc/config.scm
deco start cow-store /mnt
guix pull
guix system init /mnt/etc/config.scm /mnt
Additional information:
connection is OK, I pinged www.google.com <http://www.google.com/> after enabling enp0s3
I divided /dev/sda to two partitions:
/dev/sda1 BIOS boot, 200M
/dev/sda2 Linux filesystem, 15.8G
in /mnt/etc/config.scm I changed /dev/sdX to /dev/sda

We all know that hydra is being overloaded all the time. The guix system init process is failing on random packages due to the network issues. I am rerunning the command every time. However something strange happens at python package. Every time after downloading 15.3M the process is being killed but is it?

Here is the output:
http://hydra.gnu.org/bla/bla/hash-python-2.7.10 <http://hydra.gnu.org/bla/bla/hash-python-2.7.10> 1.3MiB/s 00:12 | 15.3MiB transferred
killing process 20890
killing process 20890: No such process

I am not returned to the command prompt after this. It is stuck that way forever. I tried it so many times already - no luck so far. Someone on #guix said that this may be the disk space issue so I created fixed disk image (rather than dynamically resizing one) and re-run the whole process. After failing with python package and running df /dev/sda2 usage is on 8% so I believe this is not the problem here.

Any ideas?

Thanks in advance.

dptd


[-- Attachment #2: Type: text/html, Size: 2939 bytes --]

Re: Initial installation - stuck at installing python (killing process)

[Tobias Geerinckx-Rice]

Hallo,

On 19/03/2016, dptdescribe <dptdescribe@gmail.com> wrote:
> We all know that hydra is being overloaded all the time. The guix system
> init process is failing on random packages due to the network issues. I am
> rerunning the command every time. However something strange happens at
> python package. Every time after downloading 15.3M the process is being
> killed but is it?

It's not you, it's Hydra.

I got the same error (also consistently at 15.3M) yesterday. I suspect
an incompletely cached file in the front end, but that's just a guess.
Maybe someone else knows.

The good news: adding a ‘--fallback’ switch to build Python locally
should do the trick. It did here. Just be prepared to wait a while on
a slower VM.

Kind regards,

T G-R

Re: Initial installation - stuck at installing python (killing process)

[dptdescribe]

[-- Attachment #1: Type: text/plain, Size: 2306 bytes --]

Thanks Tobias.

It worked for the python issue but I came back to the error which I had in the first place a few days ago. After building guix (?) it run some tests. It fails at tests/store.scm. I cannot find any logs anywhere, I asked about that on the #guix few days ago and noone could help. There are 61 tests, 2 are skipped and 1 is failed. It prints message to see ./test-suite.log but this file doesn’t exist. Actually one of the logs asks me to report a bug on the mailing list. ;)

All I did was killing the process (Ctrl-C) after seeing the well known “killing process / no such process” message after running the following command:
# guix system init /mnt/etc/config.scm /mnt

Then I re-run the system init with fallback flag:
# guix system init /mnt/etc/config.scm /mnt —fallback

As I wrote - it helped for the python issue and proceed further.

After failed test I tried to run again the first command (without —fallback) to download guix rather than building it from the source, however it didn’t helped. Maybe there is some cache which tells guix “hey - I have the source files and I already build the project, maybe we should build it again rather than download anything?” - no idea...

I cannot copy-paste the log so please see few screenshots on imgur.
http://imgur.com/a/5fsQG <http://imgur.com/a/5fsQG>

Thanks once again.

dptd

> On 19 Mar 2016, at 15:37, Tobias Geerinckx-Rice <tobias.geerinckx.rice@gmail.com> wrote:
> 
> Hallo,
> 
> On 19/03/2016, dptdescribe <dptdescribe@gmail.com> wrote:
>> We all know that hydra is being overloaded all the time. The guix system
>> init process is failing on random packages due to the network issues. I am
>> rerunning the command every time. However something strange happens at
>> python package. Every time after downloading 15.3M the process is being
>> killed but is it?
> 
> It's not you, it's Hydra.
> 
> I got the same error (also consistently at 15.3M) yesterday. I suspect
> an incompletely cached file in the front end, but that's just a guess.
> Maybe someone else knows.
> 
> The good news: adding a ‘--fallback’ switch to build Python locally
> should do the trick. It did here. Just be prepared to wait a while on
> a slower VM.
> 
> Kind regards,
> 
> T G-R


[-- Attachment #2: Type: text/html, Size: 3515 bytes --]

Re: Initial installation - stuck at installing python (killing process)

[Ludovic Courtès]

Tobias Geerinckx-Rice <tobias.geerinckx.rice@gmail.com> skribis:

> On 19/03/2016, dptdescribe <dptdescribe@gmail.com> wrote:
>> We all know that hydra is being overloaded all the time. The guix system
>> init process is failing on random packages due to the network issues. I am
>> rerunning the command every time. However something strange happens at
>> python package. Every time after downloading 15.3M the process is being
>> killed but is it?
>
> It's not you, it's Hydra.
>
> I got the same error (also consistently at 15.3M) yesterday. I suspect
> an incompletely cached file in the front end, but that's just a guess.
> Maybe someone else knows.

I don’t ;-), but could you give the exact URL that’s failing for you?
I’d be happy to investigate.

Now, I think dptdescribe is not using the mirror at all, but directly
talking to hydra.gnu.org.

dptdescribe, could you try something like:

  guix system init config.scm /mnt \
    --substitute-urls=http://mirror.hydra.gnu.org

This mirror should be more responsive than hydra.gnu.org, although
substitutes could also be missing from there.

Thanks,
Ludo’.

Re: Initial installation - stuck at installing python (killing process)

[Tobias Geerinckx-Rice]

Ludovic,

On 19/03/2016, Ludovic Courtès <ludo@gnu.org> wrote:
> Tobias Geerinckx-Rice <tobias.geerinckx.rice@gmail.com> skribis:
>
>> On 19/03/2016, dptdescribe <dptdescribe@gmail.com> wrote:
>>> We all know that hydra is being overloaded all the time. The guix system
>>> init process is failing on random packages due to the network issues. I
>>> am
>>> rerunning the command every time. However something strange happens at
>>> python package. Every time after downloading 15.3M the process is being
>>> killed but is it?
>>
>> It's not you, it's Hydra.
>>
>> I got the same error (also consistently at 15.3M) yesterday. I suspect
>> an incompletely cached file in the front end, but that's just a guess.
>> Maybe someone else knows.
>
> I don’t ;-), but could you give the exact URL that’s failing for you?
> I’d be happy to investigate.

I'm afraid it's gone to the great scrollback buffer in the sky. If it
happens again I'll post it here.

Kind regards,

T G-R

Re: Initial installation - stuck at installing python (killing process)

[dptdescribe]

Ludovic,

substitute-urls helped for the python issue, thanks. However, at the end, the whole process still fails on tests phase which I described before. :(

tests/store.scm
guix system: error: build failed: build of `/gnu/store/…db1-guix-0.9.0.71e2065.drv’ failed

dptd

> On 19 Mar 2016, at 22:16, Tobias Geerinckx-Rice <tobias.geerinckx.rice@gmail.com> wrote:
> 
> Ludovic,
> 
> On 19/03/2016, Ludovic Courtès <ludo@gnu.org> wrote:
>> Tobias Geerinckx-Rice <tobias.geerinckx.rice@gmail.com> skribis:
>> 
>>> On 19/03/2016, dptdescribe <dptdescribe@gmail.com> wrote:
>>>> We all know that hydra is being overloaded all the time. The guix system
>>>> init process is failing on random packages due to the network issues. I
>>>> am
>>>> rerunning the command every time. However something strange happens at
>>>> python package. Every time after downloading 15.3M the process is being
>>>> killed but is it?
>>> 
>>> It's not you, it's Hydra.
>>> 
>>> I got the same error (also consistently at 15.3M) yesterday. I suspect
>>> an incompletely cached file in the front end, but that's just a guess.
>>> Maybe someone else knows.
>> 
>> I don’t ;-), but could you give the exact URL that’s failing for you?
>> I’d be happy to investigate.
> 
> I'm afraid it's gone to the great scrollback buffer in the sky. If it
> happens again I'll post it here.
> 
> Kind regards,
> 
> T G-R

Re: Initial installation - stuck at installing python (killing process)

[Ludovic Courtès]

dptdescribe <dptdescribe@gmail.com> skribis:

> tests/store.scm
> guix system: error: build failed: build of `/gnu/store/…db1-guix-0.9.0.71e2065.drv’ failed

This looks like what’s reported at <http://bugs.gnu.org/23056>.

Could you answer the questions I asked there?  :-)

Thanks,
Ludo’.

Re: Initial installation - stuck at installing python (killing process)

[dptdescribe]

[-- Attachment #1: Type: text/plain, Size: 798 bytes --]

Yeap, looks like the exact same issue.

Answering your questions from the bug conversation:
1. # guix-daemon —version
guix-daemon (GNU Guix) 0.9.0
2. # df /tmp
Filesystem	1K-blocks	Used	Available	Use%	Mounted on
unionfs		2854060		830856	2023204	30%		/

On /mnt/tmp it should be ext4 because I formatted whole partition after creation (mkfs.ext4).

Thanks.

Best regards,
dptd

> On 21 Mar 2016, at 10:23, Ludovic Courtès <ludo@gnu.org> wrote:
> 
> dptdescribe <dptdescribe@gmail.com> skribis:
> 
>> tests/store.scm
>> guix system: error: build failed: build of `/gnu/store/…db1-guix-0.9.0.71e2065.drv’ failed
> 
> This looks like what’s reported at <http://bugs.gnu.org/23056>.
> 
> Could you answer the questions I asked there?  :-)
> 
> Thanks,
> Ludo’.


[-- Attachment #2: Type: text/html, Size: 2405 bytes --]

Re: Initial installation - stuck at installing python (killing process)

[dptdescribe]

[-- Attachment #1: Type: text/plain, Size: 1269 bytes --]

Hello there,
Just to let you know. Running:
# mount -t tmpfs tmp /tmp

Before guix system init helped and I have finally GuixSD up and running.

The problem was with the /tmp filesystem. The question is why and how it can be solved?

Best regards,
dptd

> On 21 Mar 2016, at 18:09, dptdescribe <dptdescribe@gmail.com> wrote:
> 
> Yeap, looks like the exact same issue.
> 
> Answering your questions from the bug conversation:
> 1. # guix-daemon —version
> guix-daemon (GNU Guix) 0.9.0
> 2. # df /tmp
> Filesystem	1K-blocks	Used	Available	Use%	Mounted on
> unionfs		2854060		830856	2023204	30%		/
> 
> On /mnt/tmp it should be ext4 because I formatted whole partition after creation (mkfs.ext4).
> 
> Thanks.
> 
> Best regards,
> dptd
> 
>> On 21 Mar 2016, at 10:23, Ludovic Courtès <ludo@gnu.org <mailto:ludo@gnu.org>> wrote:
>> 
>> dptdescribe <dptdescribe@gmail.com <mailto:dptdescribe@gmail.com>> skribis:
>> 
>>> tests/store.scm
>>> guix system: error: build failed: build of `/gnu/store/…db1-guix-0.9.0.71e2065.drv’ failed
>> 
>> This looks like what’s reported at <http://bugs.gnu.org/23056 <http://bugs.gnu.org/23056>>.
>> 
>> Could you answer the questions I asked there?  :-)
>> 
>> Thanks,
>> Ludo’.
> 


[-- Attachment #2: Type: text/html, Size: 3374 bytes --]

Re: Initial installation - stuck at installing python (killing process)

[Ludovic Courtès]

dptdescribe <dptdescribe@gmail.com> skribis:

> # mount -t tmpfs tmp /tmp
>
> Before guix system init helped and I have finally GuixSD up and running.

Good, thanks for letting us know!

> The problem was with the /tmp filesystem. The question is why and how it can be solved?

See <http://debbugs.gnu.org/cgi/bugreport.cgi?bug=23056#22> for an
explanation.

It’ll be fixed either by mounting a tmpfs on /tmp as you did, or
something else to avoid having /tmp in RAM; not sure yet.

Ludo’.