Install older package version in Alpine - alpine-linux

So recently (5th September) the Alpine Linux package repo was updated to postgresql-client 12.4
I'm referencing version 12.3 in my Dockerfile (apk add postgresql-client=~12.3). Is it not possible to install that version now?
I'd like to update on my time and terms, why should I be forced to update now? Is there another repository I can add to use the older version?
Thanks

Unfortunately, Alpine packages are always updated in place to the latest version, and older versions are discarded. This could be painful, indeed...
Usually, when a package is updated, it's updated with all Alpine distro versions that it's compatible to. For example, postgresql-client was bumped to 12.4-r0 on edge, v3.12 and v3.11, but on Alpine v3.10 repos you'll still find 11.9-r0. In case this was enough, the old version could be installed from the desired repository, as long as it lasts, using:
apk add postgresql-client=11.9-r0 --repository=http://dl-cdn.alpinelinux.org/alpine/v3.10/main
However, since 12.3 doesn't live in the official Alpine repositories anymore, you could rely on an external Docker image, instead.
Luckily, the postgres official images has version tags, and you can find the desired Alpine image for 12.3:
$ wget -q https://registry.hub.docker.com/v1/repositories/postgres/tags -O - | jq -r '.[].name' | grep 12.3
12.3
12.3-alpine
Therefore, you can use FROM:postgres:12.3-alpine to get the desired version from.
In tougher cases, where the Alpine package version is updated, and couldn't be found in other images, the only resort may be building from source.

for example; the latest dnsmasq version ins 2.84-r0 at now, if you install 2.83-r0, will:
$ docker run --rm -ti alpine:3.13
$ apk add 'dnsmasq-dnssec==2.83-r0'
fetch https://mirrors.aliyun.com/alpine/v3.13/main/x86_64/APKINDEX.tar.gz
fetch https://mirrors.aliyun.com/alpine/v3.13/community/x86_64/APKINDEX.tar.gz
ERROR: unable to select packages:
dnsmasq-dnssec-2.84-r0:
breaks: world[dnsmasq-dnssec=2.83-r0]
The best thing you can achieve is using repositories of the earlier releases, at the websiete https://pkgs.alpinelinux.org/packages to search the old version, will find 2.83-r0 in https://pkgs.alpinelinux.org/packages?name=dnsmasq&branch=v3.12.
so add the old repo
$ echo 'http://dl-cdn.alpinelinux.org/alpine/v3.12/main' >> /etc/apk/repositories
$ apk add 'dnsmasq-dnssec==2.83-r0'
fetch http://mirrors.aliyun.com/alpine/v3.12/main/x86_64/APKINDEX.tar.gz
fetch http://mirrors.aliyun.com/alpine/v3.12/community/x86_64/APKINDEX.tar.gz
(1/3) Installing gmp (6.2.0-r0)
(2/3) Installing nettle (3.5.1-r1)
(3/3) Installing dnsmasq-dnssec (2.83-r0)
Executing dnsmasq-dnssec-2.83-r0.pre-install
Executing busybox-1.31.1-r16.trigger
OK: 7 MiB in 17 packages

Another solution based on the answer of #valiano.
For upgrading postgresql to a newer version, it is recommended to use the higher version pg_dump binaries. But how to get these into your image?
This works for me:
Dockerfile:
ARG VERSION=10
ARG UPGRADE_VERSION=11
ARG TYPE
###############################################################
# Normal server
###############################################################
FROM postgres:${VERSION}-alpine AS server
RUN apk update \
&& apk add --upgrade apk-tools \
&& apk upgrade --available
COPY /rootfs/ /
###############################################################
# Upgrade version with upgrade executables
###############################################################
FROM postgres:${UPGRADE_VERSION}-alpine AS upgrade_version
RUN apk update \
&& apk add --upgrade apk-tools \
&& apk upgrade --available
###############################################################
# Add postgresql upgrade client executables to upgrade_server_layer
###############################################################
FROM server AS upgrade_server
RUN mkdir -p /usr/local/postgresql/upgrade
COPY --from=upgrade_version /usr/local/bin/pg* /usr/local/postgresql/upgrade/
###############################################################
# Final version
###############################################################
FROM ${TYPE}server AS final
Then build your normal server as:
docker build --build-arg TYPE= --build-arg VERSION=11 --build-arg UPGRADE_VERSION=12 -t my_normal_server:11 .
And a server with upgrade binaries, to make the dumpfile:
docker build --build-arg TYPE=upgrade_ --build-arg VERSION=10 --build-arg UPGRADE_VERSION=11 -t my_upgrade_server:10 .
Upgrade scenario if current version is postgresql 10 and you want to upgrade to 11:
Build an upgrade version and a normal version.
Stop the postgresql 10 container and replace it with the my_upgrade_version:10
Create a dumpfile with the /usr/local/postgresql/upgrade/pg_dump.
Create a new postgresql 11 container with my_normal_version:11 with access to the dumpfile and use pg_restore to restore the created dumpfile.

About making a dry single-point-of-entry for fixed package versions:
I use the following method, where I have an .env file where I store the PG major and minor version. I only need to update the .env-file entry with preferred version numbers and rebuild my images to upgrade the Postgres.
As long as the package is present with the corresponding version in the dockerhub and the PG version itself isn't deprecated by the alpine repositories or sth:
Step 1: Specify the PG version in .env file as single point of entry:
PG_MAJOR_VERSION=14
PG_MINOR_VERSION=5
Step 2: Reference the db-package inside the docker-compose:
services:
db:
image: postgres:${PG_MAJOR_VERSION}.${PG_MINOR_VERSION}-alpine
Step 3: Use the variables inside the Dockerfile itself if needed:
ARG RUBY_VERSION
ARG DISTRO_NAME
FROM ruby:${RUBY_VERSION}-${DISTRO_NAME}
# Need to define the args again:
ARG DISTRO_NAME
ARG PG_MAJOR_VERSION
RUN apk add --update build-base bash bash-completion libffi-dev tzdata postgresql$PG_MAJOR_VERSION-client postgresql$PG_MAJOR_VERSION-dev nodejs npm yarn
NB! The FROM clause loses the ARG-variables defined before it. Therefore if you need them later then you need to define them again after the FROM-clause. This issue is described in more detail in this Github issue
Depending on the packages you wish to install you can specify the minor version or other suffixes as needed (for ex for postgresql15-client-15.1-r0 and postgresql15-dev-15.1-r0: package etc)

Related

Postgres: Install a specific version of plv8

I'm installing plv8 in docker:
FROM postgres:13 AS build
ENV PLV8_VERSION=v3.0.0
RUN apt-get update && apt-get upgrade \
&& apt-get install -y git curl glib2.0 libc++-dev python python3-pip
libv8-dev postgresql-server-dev-$PG_MAJOR libncurses5
RUN pip install pgxnclient
RUN pgxn install plv8
This still seems to install the 2.3.11 version of plv8 though, which is incompatible with Postgres 13.
Is there any way I can specify the version that pgxn installs? Or any other way I can install a Postgres 13 version of plv8?
You can use our finished docker-images with postgres and plv8. It's free, images for Postgres 13, 14, and 15 based on Debian and Alpine are available, amd64 and arm64 architectures supported.
docker pull sibedge/postgres-plv8
Default uses Alpine and last Postgres version. All available tags here
Pay attention, that BigInt is not serializable in v8 and by default in plv8 v3.0.0 and higher BigInt numbers are converted into string(!). But if you need BigInt as numbers support, use this image with Postgres and specific version of plv8:
docker pull sibedge/postgres-plv8-bigint
OR you can use our binaries with these Dockerfiles and fast build compact images with postgres and plv8 yourself:
Postgres 14.2, plv8 v3.0.0, Alpine based. size of image is 235MB.
Postgres 13.6, plv8 v3.0.0, Alpine based. size of image is 231MB.
Postgres 13.4, plv8 v2.13.15, Debian based. size of image is 351MB.
Postgres 13.6, plv8 v3.0.0, Debian based. size of image is 427MB.
$ git clone https://github.com/sibedge-llc/plv8-build.git
$ cd plv8-build/docker
$ docker build -t pg14-plv8-3
$ docker run -it -d --name pg14-plv8-3 -e POSTGRES_USER=postgres -e POSTGRES_PASSWORD=postgres -p 5432:5432/tcp pg14-plv8-3
You can use it right now.
It looks like the latest versions of plv8 are not yet published on pgxn network yet. The last published version is 2.3.11 and that is what you see here.
To install the latest version of plv8 you can update your Dockerfile to build plv8 from source by following the build instructions. A good starting point would be to refer to the docker image clkao/postgres-plv8 which was built with postgres:10 base image.

Cannot install php7-mongodb in alpine linux

I am trying to install php7-mongodb with my dockerfile. The same worked until yesterday but today dockerbuild is throwing the following error
Step 4/12 : RUN apk add php7-mongodb
---> Running in b6713ea15c9d
php7-mongodb (missing):
ERROR: unsatisfiable constraints:
required by: world[php7-mongodb]
The command '/bin/sh -c apk add php7-mongodb' returned a non-zero code: 1
I tried with apk update and apk upgrade in the dockerfile but still not working. any idea what the issue is?
I am using alpine edge as base version.
We can recreate the issue by using the following dockerfile
FROM alpine:edge
RUN echo "http://dl-cdn.alpinelinux.org/alpine/edge/testing" >> /etc/apk/repositories
RUN apk update && apk upgrade
RUN apk add php7-mongodb
tldr: The package renamed to php7-pecl-mongodb in 3.9 and no longer exists in later releases
The package was renamed to php7-pecl-mongodb with this commit: https://git.alpinelinux.org/aports/commit/?id=b3d534d73c690baf458d9cda5dca5ee52ca9cafc
But the package was removed shortly after that due to a nonfree licence change by the package: https://git.alpinelinux.org/aports/commit/community/php7-pecl-mongodb/APKBUILD?id=8a901de31fa055ed591d487e12f8bb9ffcc0df21
According to https://pkgs.alpinelinux.org/packages?name=php7-mongodb&branch=edge there is no such package in alpine edge version.
The latest alpine version that has the php7-mongodb package in its repos seems to be 3.8: https://pkgs.alpinelinux.org/packages?name=php7-mongodb&branch=v3.8
I can't tell you what's the reason that this package is not contained in 3.9 and edge. But if this is an option for you use alpine:3.8 as the base image instead of alpine:edge
Using Alpine 3.8 will fix the issue for you. Change
FROM alpine:edge
To
FROM alpine:3.8
You will also need to change the repository URL so it's point at v3.8. I've been using main so the second line would look like below for me.
RUN echo "http://dl-cdn.alpinelinux.org/alpine/v3.8/main" >> /etc/apk/repositories
I don't think /testing exists for older releases so you'll need to experiment to find out what works for you. Or perhaps someone else can clarify this for us.
It means you're not using the latest Alpine but you will be able to build the container.

How to install VS Code in Alpine Linux

I have an operating environment that is Alpine linux only and I need to install VS Code. How can VS Code be run on Alpine Linux?
Dockerfile:
FROM node:14.19.0-alpine
RUN set -x \
&& sed -i 's/dl-cdn.alpinelinux.org/mirrors.tuna.tsinghua.edu.cn/g' /etc/apk/repositories \
&& apk upgrade \
&& apk --no-cache add alpine-sdk bash libstdc++ libc6-compat \
&& npm config set python python3
RUN git clone --recursive https://github.com/cdr/code-server.git
RUN cd code-server
RUN yarn global add code-server
ENV PASSWORD=changeme
ENTRYPOINT code-server --bind-addr 0:8443
Commands:
docker build . -t vscode
docker run -d -e PASSWORD=111111 -p8443:8443 vscode:latest
http://hostname:8443
Download it in Flatpak repos, it will run natively in a Gnome SDK environment.
Use a self-hosted environment such as Theia (https://www.theia-ide.org/index.html) or coder-editor (https://coder.com/). I've never tried them, I use the Flatpak one, but they seem interesting (you can "build" your osn editor in a Node environment).
Apologies for necrobump, but as what Marco suggested, coder.com has moved to github
the software, code-server is quite litterally VSCode, as a web application, I have been using this for about half a year and it is quite well developed, Alpine support is still spotty but i recall getting a few releases to function well a while back when i ran Alpine as my main.

Helm: Incompatible versions between client and server

After I have run helm list I got following error:
Error: incompatible versions client[v2.9.0] server[v2.8.2]
I did a helm init to install the compatible tiller version
"Warning: Tiller is already installed in the cluster.
(Use --client-only to suppress this message, or --upgrade to upgrade Tiller to the current version.)".
Any pointers?
Like the OP, I had this error:
$ helm list
Error: incompatible versions client[v2.10.0] server[v2.9.1]
Updating the server wasn't an option for me so I needed to brew install a previous version of the client. I hadn't previously installed client[v2.9.1] (or any previous client version) and thus couldn't just brew switch kubernetes-helm 2.9.1. I ended up having to follow the steps in this SO answer: https://stackoverflow.com/a/17757092/2356383
Which basically says
Look on Github for the correct kubernetes-helm.rb file for the version you want (2.9.1 in my case): https://github.com/Homebrew/homebrew-core/search?q=kubernetes-helm&type=Commits
Click the commit hash (78d6425 in my case)
Click the "View" button to see the whole file
Click the "Raw" button
And copy the url: https://raw.githubusercontent.com/Homebrew/homebrew-core/78d64252f30a12b6f4b3ce29686ab5e262eea812/Formula/kubernetes-helm.rb
Now that I had the url for the correct kubernetes-helm.rb file, I ran the following:
$ brew unlink kubernetes-helm
$ brew install https://raw.githubusercontent.com/Homebrew/homebrew-core/78d64252f30a12b6f4b3ce29686ab5e262eea812/Formula/kubernetes-helm.rb
$ brew switch kubernetes-helm 2.9.1
Hope this helps someone.
To upgrade your tiller version to the same version of the client, just run helm init --upgrade
NOTE: If you're trying to downgrade the server version to match your local client version, run the following instead:
helm init --upgrade --force-upgrade
Another alternative, if changing the server version is not an option, is to use the
helm installer script
The script lets you chose a specific version like so
./get_helm.sh -v v2.13.1
Another approach to using different versions through Docker.
https://hub.docker.com/r/alpine/helm
Example: list helm packages installed
docker run -it --rm \
-v ~/.kube/config:/root/.kube/config \
-v ~/.helm:/root/.helm alpine/helm:2.9.1 \
list
This is a long command; but it can be shortened with an alias
alias helm_2_9_1="docker run -ti --rm \
-v $(pwd):/apps -v ~/.kube/config:/root/.kube/config \
-v ~/.helm:/root/.helm alpine/helm:2.9.1"
And then the command is
helm_2_9_1 list
This answer is for who want to choose(downgrade) helm client version, and the brew install is not work.You can just manually install the binary file from here.
example:
you can unlink the current helm
brew unlink kubernetes-helm
choose and download the helm version you want in github helm------v2.8.2
unzip the file and put the helm unix executable binary file into /usr/local/bin directory
go to the directory you just downloaded
cd /Users/your_name/Downloads
unzip the file
gunzip -c helm-v2.8.2-darwin-amd64.tar.gz | tar xopf -
copy to the bin directory
cp darwin-amd64/helm /usr/local/bin
now you will see the right version of helm you want
helm version
For those having installed their helm client with snap, to downgrade/upgrade it to a specific version you can simply:
Uninstall it: snap remove helm
Check the available versions: snap info helm
Install the one you want: snap install helm --channel=X.X/stable --classic
This probably isn't the most advanced answer... but my team runs kubernetes clusters that already have tiller installed. While setting up a new laptop, I wanted my helm to match the tiller version, so I found it like this:
TILLER_POD=`kubectl get pods -n kube-system | grep tiller | awk '{print $1}'`
kubectl exec -n kube-system $TILLER_POD -- /tiller -version
Then I just used the normal helm install instructions from that release number (being on Linux, its basically just curl and unzip to /usr/local/bin).
If you are windows user and installed helm through choco, firstly go its folder (mine is C:\ProgramData\chocolatey) and delete helm.exe from bin folder.
Then, corresponding heml.exe file should be downloaded. By using the above comments, decide the location where you will download exe from. For instance, I used that path: https://get.helm.sh/helm-v2.14.3-windows-amd64.tar.gz
Finally extract the helm.exe from tar and move into choco bin folder. Of course, you can directly add this exe into the path.
I experienced same issue, but in my case I wanna only to upgrade Tiller to specific version (because helm client is running remotely).
So, error was:
Error: UPGRADE FAILED: incompatible versions client[v2.11.0] server[v2.9.1]
Accordingly to documentation I've run:
$ kubectl --namespace=kube-system set image deployments/tiller-deploy tiller=gcr.io/kubernetes-helm/tiller:v2.11.0
deployment.extensions/tiller-deploy image updated
Documentation reference:
https://helm.sh/docs/install/#upgrading-tiller

how to install package from alpine aports

I have been trying to install a package that exists in alpine aports
and specifically that one but I cannot find how. Is it even possible? If yes, how?
The package filebeat you defined in the question is located in edge branch of testing repository. There is no such repo in alpine container by default.
In order to install a filebeat package on the alpine platform we need:
1. Add testing repo:
/ # echo "http://dl-cdn.alpinelinux.org/alpine/edge/testing/" >> /etc/apk/repositories
2. Install the filebeat package:
/ # apk add --no-cache filebeat
fetch http://dl-cdn.alpinelinux.org/alpine/v3.7/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.7/community/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/edge/testing/x86_64/APKINDEX.tar.gz
WARNING: This apk-tools is OLD! Some packages might not function properly.
(1/1) Installing filebeat (5.6.3-r0)
Executing busybox-1.27.2-r7.trigger
OK: 20 MiB in 12 packages