Top is not working when starting using docker

[strav66]

Docker postgrespro-standard work in centos 8.3

pgcenter 0.6.6 work in docker

docker exec -it srv-db-APSD-VO pgcenter --version
pgcenter v0.6.6 4f87424-master
docker exec -it srv-db-APSD-VO pgcenter top

pgcenter 0.7.0 not work docker:

docker exec -it srv-db-APSD-VO /opt/pgpro/data/pgcenter --version
pgcenter v0.7.0 45c8bf0d0098f377692be886691eba9ee6005808-release
docker exec -it srv-db-APSD-VO /opt/pgpro/data/pgcenter top -h localhost
panic: send on closed channel

goroutine 65 [running]:
github.com/lesovsky/pgcenter/top.collectStat(0x9e5620, 0xc000138c40, 0xc000122080, 0xc000382000, 0xc0001281e0)
        /home/runner/work/pgcenter/pgcenter/top/stat.go:57 +0x371
github.com/lesovsky/pgcenter/top.doWork.func1(0x9e5620, 0xc000138c40, 0xc000148000, 0xc000382000, 0xc000380000)
        /home/runner/work/pgcenter/pgcenter/top/ui.go:77 +0x65
created by github.com/lesovsky/pgcenter/top.doWork
        /home/runner/work/pgcenter/pgcenter/top/ui.go:76 +0xeb

[lesovsky]

Hi, thanks for report,
What docker image is used?

[strav66]

my build docker

Dockerfile.zip

[lesovsky]

What happen if run container and then run pgcenter being inside?
Like:

docker run --rm -ti IMAGE /bin/bash
pgcenter top ...

[strav66]

Работает, но странно:
/opt/pgpro/data/pgcenter top -h localhost

нет статистики: процессор, память и т.д.

Если запускать так (внутри docker):
/opt/pgpro/data/pgcenter top

тогда ругается:
failed to connect to host=/var/run/postgresql user=postgres database=: dial error (dial unix /var/run/postgresql/.s.PGSQL.5432: connect: no such file or directory)

для 0.6.6 соответственно:
pgcenter top -h localhost

pgcenter top

Всё отлично и со статистикой

[lesovsky]

I tested various cases using different docker images and bug reproduced in most of cases. Sometimes pgcenter starts as expected but at next, the same attempt it crashed again (even 0.6.6 might hang with black screen). I suspect there something related to UI processing when docker starts, but need to dig deeper.
I changed the title of the issue to reflect the problem more precisely.

Also I strongly recommended to avoid using national languages and post comments using English.

[strav66]

OK

tried 10 times sequentially run pgcenter 0.7.0:
docker exec -it srv-db-APSD-VO /opt/pgpro/data/pgcenter top -h localhost
6 times started with errors, 4 times - ok (but without CPU statistics, etc.)

error:
panic: send on closed channel

goroutine 53 [running]:
github.com/lesovsky/pgcenter/top.collectStat(0x9e5620, 0xc000310040, 0xc000122080, 0xc000326000, 0xc0001281e0)
/home/runner/work/pgcenter/pgcenter/top/stat.go:57 +0x371
github.com/lesovsky/pgcenter/top.doWork.func1(0x9e5620, 0xc000310040, 0xc000140000, 0xc000326000, 0xc000318010)
/home/runner/work/pgcenter/pgcenter/top/ui.go:77 +0x65
created by github.com/lesovsky/pgcenter/top.doWork
/home/runner/work/pgcenter/pgcenter/top/ui.go:76 +0xeb

pgcentert 0.6.6 - never broke (maybe I was just lucky)

[lesovsky]

Hi,
check out new v0.8.0 release, both issues related to absence of stats and panic are fixed. Waiting for testing and feedback.

[strav66]

pgcenter 0.8.0 works fine!

[strav66]

Addition:
when exiting (closing pgcenter) an error occurs:

panic: runtime error: slice bounds out of range [:-1]

goroutine 49 [running]:
github.com/nsf/termbox-go.Init.func1()
        /home/runner/go/pkg/mod/github.com/nsf/termbox-go@v0.0.0-20180819125858-b66b20ab708e/api.go:95 +0x36a
created by github.com/nsf/termbox-go.Init
        /home/runner/go/pkg/mod/github.com/nsf/termbox-go@v0.0.0-20180819125858-b66b20ab708e/api.go:84 +0x730

[lesovsky]

Panic occurs at every time or spontaneously? What docker image do you use?
Try to reproduce error using lesovsky/pgcenter:v0.8.0 image.

[strav66]

Panic occurs at every time or spontaneously?

spontaneously

What docker image do you use?

my docker image

[lesovsky]

What about attempt to reproduce using lesovsky/pgcenter:v0.8.0 image?

my docker image

It is insufficient information, provide more details... base image, full command definition of how pgcenter was started (including docker opts).

[strav66]

What about attempt to reproduce using lesovsky/pgcenter:v0.8.0 image?

tried lesovsky / pgcenter: v0.8.0 image
The error was not reproduced

but it is not possible to use the name localhost with it!
I think the error is related to localhost

my docker image

#87 (comment)

It is insufficient information, provide more details... base image, full command definition of how pgcenter was started (including docker opts).

docker runs via docker-compose (docker-compose.yml.tar.gz):
docker-compose up -d

pgcenter upload volume PGDATA
run pgcenter and connect to a running docker with the command:

docker exec -it srv-db1C-ERP /opt/pgpro/data/pgcenter top -h localhost

after a few minutes (3-4), sometimes an error occurs (pgcenter continues to work!):

kernel:unregister_netdevice: waiting for lo to become free. Usage count = 1

If it is ok to exit pgcenter (with the q key) the message appears:

panic: runtime error: slice bounds out of range [:-1]

goroutine 49 [running]:
github.com/nsf/termbox-go.Init.func1()
        /home/runner/go/pkg/mod/github.com/nsf/termbox-go@v0.0.0-20180819125858-b66b20ab708e/api.go:95 +0x36a
created by github.com/nsf/termbox-go.Init
        /home/runner/go/pkg/mod/github.com/nsf/termbox-go@v0.0.0-20180819125858-b66b20ab708e/api.go:84 +0x730

[lesovsky]

Trying to build the image, need the values of variables from docker-compose:

    build:
      context: ../docker/postgrespro-standard

      dockerfile: Dockerfile.${PG_Version}
      args: 
        - SERVER_VERSION=${PG_Version}
        - SERVER_VERSION_DOT=${PG_Version_dot}
        - KEYS_DOT=${rpm_keys_dot}

No need SERVER_VERSION, but need other values of SERVER_VERSION_DOT, KEYS_DOT and maybe content of ../docker/postgrespro-standard (but not sure).

[strav66]

If you carefully look at the attached archive (docker-compose.yml.tar.gz), you can see the .env file in which the necessary variables are registered

[lesovsky]

Oh, hidden file ))) yep, found now

[lesovsky]

Both ways failed due to necessary files are not exist.

1. using docker-compose build

docker-compose build
ERROR: build path /tmp/docker/postgrespro-standard either does not exist, is not accessible, or is not a valid URL

2. manual docker build also fails

docker build --build-arg SERVER_VERSION=12 --build-arg SERVER_VERSION_DOT=6.1 --build-arg KEYS_DOT=0.3 -t local/pg1c:dev -f Dockerfile.10 .
... too many lines ...
COPY failed: file not found in build context or excluded by .dockerignore: stat container/docker-entrypoint.sh: file does not exist

Also such files are not found in attached archives. Maybe you just build and publish the image on DockerHub?

[strav66]

Complete set of everything you need for build image
docker-compose-build.tar.gz

Unpack it into some folder
go to the ERP folder
we give the command:

docker-compose build

Profit

[lesovsky]

Thanks, build successful.

Upd. I've just updated pgcenter download links in Dockerfiles, from 0.6.x to 0.8.0.

[lesovsky]

Spent a quite lot of time for testing and could not reproduce the panic at quit. Also I have no idea how to catch this using unit-tests.

Would like to clarify, are you sure you build the image with proper version of pgcenter (v0.8.0)? Because I manually changed pgcenter's version in your files before build. Maybe you catch error with old version (the similarly as it was here)?