Prevent the nodejs applications restarting on every Chef run #5

New Issue

greg · 2019-01-02T22:16:42Z

greg commented

2019-01-02 22:16:42 +00:00

It is the application_npm_install resources that are causing the nodejs services to be restarted on every Chef run

I have reported the issue upstream: https://github.com/poise/poise-javascript/issues/9 but no response. We need to find a way to use this unreleased version that fixes it (it's distributed as a gem that's a dependency in the application_javascript cookbook)

It is the `application_npm_install` resources that are causing the nodejs services to be restarted on every Chef run I have reported the issue upstream: https://github.com/poise/poise-javascript/issues/9 but no response. We need to find a way to use this unreleased version that fixes it (it's distributed as a gem that's a dependency in the application_javascript cookbook)

raucao commented

2019-05-28 10:54:17 +00:00

It would seem that poise-javascript is dead now.

It would seem that `poise-javascript` is dead now.

greg commented

2019-06-04 13:43:08 +00:00

All the application/poise projects are dead, the maintainer isn't working on them anymore. I'm currently taking a look at Chef's new app build/deployment/management system, Habitat, that looks very interesting: https://www.habitat.sh

greg commented

2019-06-21 14:15:38 +00:00

I have started from the unreleased master branch of poise-javascript (rake chef:build). That includes a change that supports the new behavior of npm install, writing "up to date" to stdout when no changes have been performed.

However, npm install now runs security auditing by default, even with the --production switch, so even the newer version of poise-javascript still causes the service to restart

audited 1654 packages in 3.93s
found 4 vulnerabilities (2 moderate, 2 high)
  run `npm audit fix` to fix them, or `npm audit` for details

I have then added the --no-audit switch to prevent the service from restart at every run, caused by the output of the npm install command missing the "up to date" string, and pushed that to a new repo: d85078fe59.

I'm going to do a PR to use that repo in our Berksfile, that works great in a VM

I have started from the unreleased master branch of poise-javascript (`rake chef:build`). That includes a change that supports the new behavior of `npm install`, writing "up to date" to stdout when no changes have been performed. However, `npm install` now runs security auditing by default, even with the `--production` switch, so even the newer version of poise-javascript still causes the service to restart ``` audited 1654 packages in 3.93s found 4 vulnerabilities (2 moderate, 2 high) run `npm audit fix` to fix them, or `npm audit` for details ``` I have then added the `--no-audit` switch to prevent the service from restart at every run, caused by the output of the `npm install` command missing the "up to date" string, and pushed that to a new repo: https://github.com/67P/poise-javascript/commit/d85078fe59bd4f16d05a9ffe6e0fc449015e4440. I'm going to do a PR to use that repo in our Berksfile, that works great in a VM

greg referenced this issue from a commit

2019-06-21 15:41:05 +00:00

Switch to our own fork of the poise-javascript cookbook Includes passing the `--no-audit` switch to `npm install` to prevent the `npm_install` resource to consider the command resulted in changes, causing the service to restart at every Chef run. npm returns the expected string "up to date" when not performing the security audit Closes #5

raucao closed this issue

2019-07-02 18:46:49 +00:00

raucao reopened this issue

2019-07-03 13:57:55 +00:00

raucao commented

2019-07-03 13:58:26 +00:00

This happened again just now. I changed an env var for hal8000, but botka still restarted.

greg commented

2019-07-04 06:57:28 +00:00

It can be any of the resources inside of the application resource, I'm checking what's causing it

greg commented

2019-07-04 09:35:15 +00:00

I can't reproduce the issue in a VM, but I can confirm it still happens on barnard. Subresources of the application resource aren't updating, but still the service restarts. Logs:

Recipe: kosmos-hubot::hal8000_xmpp
  * service[hal8000_xmpp] action restart
    - restart service service[hal8000_xmpp]
  * application[/opt/hal8000_xmpp] action restart
    * service[hal8000_xmpp] action restart
      - restart service service[hal8000_xmpp]
     (up to date)
Recipe: kredits-github::default
  * application[/opt/kredits-github] action restart
    * service[kredits-github] action restart
      - restart service service[kredits-github]
     (up to date)
Recipe: kosmos-hubot::botka_freenode
  * application[/opt/botka_freenode] action restart
    * service[botka_freenode] action restart
      - restart service service[botka_freenode]
     (up to date)
Recipe: kosmos-hubot::hal8000
  * application[/opt/hal8000] action restart
    * service[hal8000_nodejs] action restart
      - restart service service[hal8000_nodejs]
     (up to date)
Recipe: sockethub::default
  * application[/opt/sockethub] action restart
    * service[sockethub_nodejs] action restart
      - restart service service[sockethub_nodejs]
     (up to date)

Not high priority, to investigate more later

I can't reproduce the issue in a VM, but I can confirm it still happens on barnard. Subresources of the application resource aren't updating, but still the service restarts. Logs: ``` Recipe: kosmos-hubot::hal8000_xmpp * service[hal8000_xmpp] action restart - restart service service[hal8000_xmpp] * application[/opt/hal8000_xmpp] action restart * service[hal8000_xmpp] action restart - restart service service[hal8000_xmpp] (up to date) Recipe: kredits-github::default * application[/opt/kredits-github] action restart * service[kredits-github] action restart - restart service service[kredits-github] (up to date) Recipe: kosmos-hubot::botka_freenode * application[/opt/botka_freenode] action restart * service[botka_freenode] action restart - restart service service[botka_freenode] (up to date) Recipe: kosmos-hubot::hal8000 * application[/opt/hal8000] action restart * service[hal8000_nodejs] action restart - restart service service[hal8000_nodejs] (up to date) Recipe: sockethub::default * application[/opt/sockethub] action restart * service[sockethub_nodejs] action restart - restart service service[sockethub_nodejs] (up to date) ``` Not high priority, to investigate more later

greg commented

2019-10-10 10:50:19 +00:00

This is not an issue anymore, confirmed after working on #104 and running Chef multiple times

greg closed this issue

2019-10-10 10:50:19 +00:00

Sign in to join this conversation.

No Milestone

No project

No Assignees

2 Participants

Notifications

Due Date

The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: kosmos/chef#5