forked from mirrors/gecko-dev
0e4ed25c34
When a task gets too routes to be indexed by the queue, we create a new dependent task that generates the index tasks. This task didn't set the index rank, so if a task gets new routes that push it over the limit, it would stop updating the old index, as it would be using an index of 0. Differential Revision: https://phabricator.services.mozilla.com/D10016 --HG-- extra : moz-landing-system : lando |
||
|---|---|---|
| .. | ||
| .eslintrc.js | ||
| Dockerfile | ||
| insert-indexes.js | ||
| npm-shrinkwrap.json | ||
| package.json | ||
| README | ||
Index-Image
===========
This image is designed to be used for indexing other tasks. It takes a task
definition as follows:
```js
{
...,
scopes: [
'index:insert-task:my-index.namespace',
'index:insert-task:...',
],
payload: {
image: '...',
env: {
TARGET_TASKID: '<taskId-to-be-indexed>',
},
command: [
'insert-indexes.js',
'my-index.namespace.one',
'my-index.namespace.two',
'....',
],
features: {
taskclusterProxy: true,
},
maxRunTime: 600,
},
}
```
As can be seen the `taskId` to be indexed is given by the environment variable
`TARGET_TASKID` and the `command` arguments specifies namespaces that it must
be index under. It is **important** to also include scopes on the form
`index:insert-task:<...>` for all namespaces `<...>` given as `command`
arguments.