Posts made by philippe.camelio_3885

For the modules, 2 ways

1. Create a second role which depend on the previous one and use
   Ensure-psmodule function to add the required module

2. You don t need the set psgallery to be trusted if you use the Force in the Ensure-psmodule
   Cf https://docs.inedo.com/docs/otter-reference-operations-powershell-ensure-powershell-module

@Justinvolved

Make a role with this desired configuration

Ensure-PsRepository
    (
        Exists: true,
        InstallationPolicy: trusted,
        SourceLocation: https://www.powershellgallery.com/api/v2,
        Name: PSGallery
    );

Assign the role to the server
Remediate

Et voilà
This should work

Cheers
PhilippeC.

Hello @atripp
You save my day.
Increasing the timeout resolve the pb.

Thank you

Hi @atripp

Thanks you for the tips

Beste reagrds
PhilippeC.

Hi
We are using Ensure-PsRepository to remove PSGallery from our server (at least for inedo agent).
This is working fine.
Role : os-w-base

    Ensure-PsRepository
    (
        Exists: false,
        InstallationPolicy: Untrusted,
        SourceLocation: https://www.powershellgallery.com/api/v2,
        Name: PSGallery
    );

As there is a mess with PSGallery some Powershell modules cannot be installed thru Proget (used as proxy for PSGallery).

So, I decided to reconfigure PSGallery using Ensure-PsRepository to be able to install theses modules.

Role : os-w-base

    Ensure-PsRepository
    (
        Exists: true,
        InstallationPolicy: Untrusted,
        SourceLocation: https://www.powershellgallery.com/api/v2,
        Name: PSGallery
    );

I try a remediation on a server (w2022), and the result is killing me

is it possible you update the Ensure-PsRepository function ?

Cheers
PhilippeC.

Hum, difficult to reproduce.
it happens with new server.
I will see if I can make a simple reproducible test case

I made a small test with Proget (Frre) on Windows - last version.
Same result

Hello @atripp
No, the older one Version 3.0.24

Hi @atripp

Thanks for your reply.
Therefore even if I write the complete name of the module - VMware.PowerCLI - , it can't be found.

I am using docker version of Proget, is it possible the cause of the issue ?

Is there anyone else using the Docker version ?

Hello,

Just to report it takes a long time (> 2 mn) to change the status of agent and assigned server roles after a successfull remediation

This delay is annoying, previous versions did not have it.
I am using docker version (2023.3)

Any suggestion to improve this behaviour if it is not a bug ?

Best regards
PhilippeC.

Hi,

I have a Feed called "public-psm" for PowerShell Gallery.
The v2023.27 is broken (previous version was too). Some modules are not visible - exemple with VMWARE.PowerCLI
This was working fine on v5.3.38

Exemple :
v2023.27

v5.3.38

the feeds were created as simple as possible
v2023.27 have no filters.

Any advice is welcome

Best Regards
PhilippeC.

Hello

Is it possible to have the lists in an alphabetic order ?
otherwise it is a little bit messy

I am using BM Version 2023.8 (Build 2) - docker

Best Regards
PhilippeC.

@atripp

Hello

The list of Upcoming & Recuring jobs is not correct.
It display old informations job which are no longer relevant.
Exemple :

When I click on the job, I have this error
.

Same error for all the lines ...

I guess some table should to be cleanup in the database ....

I am using Otter 2023.3 (Build 1) Docker version

Best Regards
PhilippeC.

@atripp
Thanks for your reply,

I am using Resource pool in this way, at the PipelineStage,
Illustration with one of our appli (angular based)

In this application, an angular apps is compiled (step 1) and deployed to a group of servers (step 2).

That's what I thought, I tried...
I may have planted a seed and maybe in the future this option will exist

I wish you all a happy holiday season

Best regards
PhilippeC.

Hello Inedo team
I have a small question.
I am using resource pool to build docker or for compilation.

I defined two server as "inedo runner" for availability and I am able to build containers or compile apps twice at the time for several aplications.
But now I need more "runner", as we have more applications.
I was thinking if it is possible to assign a number of concurrent tasks for a resource pool server in order to reduce the waiting time for a free resource.
This would reduce the number of "runner" and a neat option ... at least for me
Creating new server is an option, but I would prefer not.

Cheers
PhilippeC.

@rhessinger

[UPDATE]
The bug can be overcome using $trim function

Hello

I may have found a nice bug

From i small pipeline, I ran a test script which only display 2 variables.
The goal is to concatenate the value of these values to a third one like a docker container and "version"

I assign values at the Build step

• One from a text field
• One from a
  ** a text
  ** a list

And I display the results,
${VAR1}:${VAR2}

If VAR2 is select from a text, I have got what expected

If I select VAR2 from a list :

The carriage return is added and mess up the final variable

Here are the Pipeline used and the script

{
  "Name": "TEST",
  "Description": "Just creates a blank pipeline where that can add stages, approvals, etc., to as you need.",
  "Stages": [
    {
      "Name": "STEP 1",
      "Description": "",
      "Targets": [
        {
          "ScriptId": "TEST",
          "ServerNames": [
            "LOCALHOST"
          ],
          "ServerRoleNames": [],
          "DefaultServerContext": 1
        }
      ],
      "Gate": {
        "UserApprovals": [],
        "GroupApprovals": [],
        "DeploymentWindows": [],
        "AutomatedChecks": {
          "Issues": {
            "Statuses": []
          }
        },
        "AutomaticApprovals": []
      },
      "AllowMultipleActiveBuilds": false
    },
    {
      "Name": "STEP 2",
      "Description": "",
      "Targets": [],
      "Gate": {
        "UserApprovals": [],
        "GroupApprovals": [],
        "DeploymentWindows": [],
        "AutomatedChecks": {
          "Issues": {
            "Statuses": []
          }
        },
        "AutomaticApprovals": []
      },
      "AllowMultipleActiveBuilds": false
    }
  ],
  "TemplateVariables": [
    {
      "Name": "VAR1",
      "Description": "",
      "InitialValue": "list1",
      "Required": true,
      "Type": "List",
      "Usage": "Build",
      "PipelineStageNames": [],
      "ListValues": [
        "list1\r",
        "list2"
      ]
    },
    {
      "Name": "VAR2",
      "Description": "",
      "InitialValue": "var2",
      "Required": true,
      "Type": "Text",
      "Usage": "Build",
      "PipelineStageNames": [],
      "ListValues": []
    }
  ],
  "EventListeners": [
    {
      "Event": 5,
      "PipelineStageNames": [],
      "Type": "DeployReleaseEventListener"
    },
    {
      "Event": 5,
      "PipelineStageNames": [],
      "Type": "CreateReleaseEventListener"
    }
  ]
}

Script TEST

Log-Information ${VAR1};
Log-Information ${VAR2};
Log-Information ${VAR1}:${VAR2};

Best tests
PhilippeC.

@dean-houston
About the missing files (FileNotFoundException), I made a mistake in my docker-compose file, so the packages volume wasn't persistent.
Noob error , sorry about that.

Best regards
PhilippeC.

Hi @stevedennis
I was just looking to configure Proget as good as possible.
This part was a little bit fogus for me
I thought this was a bug ...

Thanks for your reply
Best regards
PhilippeC.

I tried to clean up:

• Remove everything
• Disable PGVC
• Enable PVGC
• Add Vulnerability Source named My PVGC
  I have this :
  
  I can't remove the extra PVGC

Oups, I inserted the wrong screenshot

If I remove the empty one, I have this:

If add a new PVGC, I got this

I undestand I have to remove OSS Index

Thanks
Philippe

Hi Inedo's team
I have PVGC twice

If I removed one, both disappear.
If I add it again, I have it twice

Any advice is welcome.
Best regards
PhilippeC.

Hi @dean-houston
The server crashed because lack of ram.
With 32 Go, no more problem

I think I have to update my 3 docker-compose to add RAM limitations for each SQL Server and Proget/Buildmaster/Otter to contain them.
Until now, I had small applications (mainly apache/mariadb) and it is the first time I have SQL Express and dotNet apps on docker.

About the logs, I did not have time to send you something relevant and helpful. I will see what I can get.

And you are right, until now I solved symptoms (adding RAM) and not the origin of the problem (unknown)

Thanks for your time s
Best regards
Philippe

Hi @atripp
I made some tests.

I increased RAM up to 32 Go with 8 vCPU (3.30 GHz)
VM located on SSD Storage

Test case:

• delete the existing npm Feed
• create a npm feed (with default parameters)
• run a npm ci from Buildmaster. I am using one of our application (npm registry set to proget)
• Assign Web.ConcurrentRequestLimit
  a) set value to 0
  b) set value to 300

My observations:

1. Server crashed with less than 32 Go of RAM.

2. With Web.ConcurrentRequestLimit to 0,
   compilation is in error due to missing packages.
   Ram peak up to 28 Go

3. With Web.ConcurrentRequestLimit to 300
   Compilation OK
   less ram consuming

Figure 1: Ram consuming for Web.ConcurrentRequestLimit

To be complete:
Proget is running as docker (docker-compose)
On the same VM, there is 2 other containers

• Buildmaster
• Otter

@philippe-camelio_3885

I will try a lower value of Web.ConcurrentRequestLimit as you siggest.

About hardware, proget vms are on the same model/config of esx server. So they should used the same amount of cpu.
I will see if I can make more tests.

Thanks for the reply

Hi @atripp

I made a test with empty npm feed.
The proget is the docker version.

Hi,

We are looking to migrate our Proget from 5.3.38 to 2023.22.
We have a NPM feed (proxy to https://registry.npmjs.org) and we have a terrible loss of performance using the new Proget - especially during the first compilation. We never saw this behaviour with Proget 5.3

Furthermore, the compilation fails with an error as some packages are not available (FileNotFoundException) while the package have been cached on Proget 2023.22.
Removing a package from the cache solved the problem
But a lot of packages have problems

The same compilation is OK when using the old Proget 5.3.38

Any thoughts ?

New proget is on Docker

Best Regards
PhilippeC.

@dean-houston

Go to Administration

Global Components

Variables
Select a scope
Click on Edit as JSON

Only variables from Scope Global are displayed as json whatever the scope you select

Hi Dean

Thanks for your reply. It was just a question, finally useless.
And Thanks you for your time

Best regards
Philippe

hi dean
Good french but poor english sorry about that

In an application, variables can be set as text, checkbox or dropdown list and be select when you start a new release.
Is it possible to do so at the group applications level or system level and use it in several applications. Just to avoid to duplicate the variable.

I would like to define a variable with a type as list at the group level or system level ?
Actually, I can do it at the release level for an application.

Any ideas ?
Best regards

Hi,
Everything is in the title ...
Whatever the variables you select, you have the system variables in a json form

Buildmaster version: docker 2023.5 (Build 4)

Best regards
Philippe

I finally found how to handle properly (I guess) the variable
The Dockerfile looks like:

ARG SOME_VAR=default_var

FROM some_image:some_version

WORKDIR /usr/src/app
RUN ....

and the Deploy script like :

Git::Checkout-Code();
Docker::Build-Image
(
    DockerfileVariables: %(SOME_VAR:myvar),
    Tag: $ReleaseNumber-pre.$BuildNumber,
    RemoveAfterPush: true
);

bur I have signing problem to push container in proget. => I start a new thread

Hi @rhessinger
I will dig in this direction.
I let you know.

Thanks for the tip.
Best regards

Hi guys,

I am looking for some help. I am trying to build a container using BM, and can't figure out how to pass env file to docker build

I am using a simple git repo which contains:

• Dockerfile
• myapp.env

How to config Docker::Build-Image to use myapp.env

I am using a very simple script :

Git::Checkout-Code();
Docker::Build-Image();

Any advice or sample is welcome

Hi @rhessinger
Pb solved,
Thanks for your fast correction

Hi,
I migrate from 2022.14 to 2023 and SSH connection is broken

Unhandled exception: The type initializer for 'Inedo.Agents.Ssh.Libssh2.NativeMethods' threw an exception.
Details:System.TypeInitializationException: The type initializer for 'Inedo.Agents.Ssh.Libssh2.NativeMethods' threw an exception.
---> System.DllNotFoundException: Unable to load shared library 'libssh2' or one of its dependencies. In order to help diagnose loading problems, consider setting the LD_DEBUG environment variable: liblibssh2: cannot open shared object file: No such file or directory
at Inedo.Agents.Ssh.Libssh2.NativeMethods.libssh2_init(Int32 flags)
at Inedo.Agents.Ssh.Libssh2.NativeMethods..cctor()
--- End of inner exception stack trace ---
at Inedo.Agents.Ssh.Libssh2.NativeMethods.libssh2_session_init_ex(IntPtr myalloc, IntPtr myfree, IntPtr myrealloc, IntPtr context)
at Inedo.Agents.Ssh.Libssh2.Libssh2Session..ctor()

I rollback and SSH is OK.

I am using the docker image tagged 23

Best regards

Hi @atripp

I may have found the pb:

The list contains an extra character \r

 "ListValues": [
    "%(env:INT,site:www1,serveur:VMXXXXXX)\r",
    "%(env:DEV,site:www2,serveur:VMXXXXXX)\r",
    "%(env:PRD,site:www3r,serveur:VMXXXXXX)\r",
    "%(env:CLO,site:www4,serveur:VMXXXXXX)\r",
    "%(env:SEC,site:www5,serveur:VMXXXXXX)"
  ],

I removed it, and now it is working fine.

      "ListValues": [
         "%(env:INT,site:www1,serveur:VMXXXXXX)",
         "%(env:DEV,site:www2,serveur:VMXXXXXX)",
         "%(env:PRD,site:www3r,serveur:VMXXXXXX)",
         "%(env:CLO,site:www4,serveur:VMXXXXXX)",
         "%(env:SEC,site:www5,serveur:VMXXXXXX)"
       ],

Problem comes from the editor of the job.
I made few tests, add a new value, remove one and the extra character is always added when I save the job

At least, the pb can be overcome.
Having access to the raw content of the Raft is very usefull,

Thanks for your time

Best regards

@gdivis
Thank you

Hello

I have problems to make scheduled jobs running in v 2023.1.

Create a small pipeline with 2 stages

• clone
• secours

Assign variables

• VAR1 = var1 (same value for both stages)
• VAR2
  • stage clone : var2-clone
  • stage secours : var2-secours

Pipeline Deployement targets :
run test otter script

Log-Information $VAR1 $VAR2;

When I create the scheduled job I have this error message :

Logged:09/19/2023 12:10:00
Level:Error
Category:Automatic Builds
Message:Unhandled exception: Failed to compare two elements in the array.
Details:System.InvalidOperationException: Failed to compare two elements in the array.
---> System.ArgumentException: Object must be of type String.
at System.String.CompareTo(Object value)
at System.Collections.Generic.ArraySortHelper`1.BinarySearch(T[] array, Int32 index, Int32 length, T value, IComparer`1 comparer)
--- End of inner exception stack trace ---
at System.Collections.Generic.ArraySortHelper`1.BinarySearch(T[] array, Int32 index, Int32 length, T value, IComparer`1 comparer)
at System.Collections.Generic.SortedList`2.IndexOfKey(TKey key)
at Inedo.TaskRunner.RunSubtaskInternalAsync[TTaskId](TTaskId id, Func`2 runAsync, String description, Object logContext, CancellationToken cancellationToken, Boolean background)

This was working in previous version

BM version : 2023.1 (docker)

Any advice is welcome

Best regards
Philippe

Hi @gdivis
Thanks for the update
I moved to docker recently - much more flexible than Microsoft

Hi Julien,
Yes, it is the manual install they are talking about.
You have to go to my.inedo.com to get the offline installers.

Best regards
PhilippeC.

Hello Julien

You can look here https://forums.inedo.com/topic/2549/cannot-add-the-principal-nt-authority-network-service-because-it-does-not-exist-or-you-do-not-have-permission?_=1693918947116

but I moved to docker since
It is a much straightforward way to deploy Proget

I had a lot of trouble in the past with this kind of configuration (French OS) sometimes it breaks the update :(

Best regards
PhilippeC.

@atripp
Good to know
Thanks