Commit 86251e74 authored by Shinya Maeda's avatar Shinya Maeda

Fix documents

parent f930e943
......@@ -77,10 +77,10 @@ cloud-native, for example on Kubernetes.
The data flow is the same as described in the [data flow section](#data-flow)
with one change: _the stored path of the first two phases is different_. This new live
trace architecture stores chunks of traces in Redis and the database instead of
trace architecture stores chunks of traces in Redis and a persistent store (object storage or database) instead of
file storage. Redis is used as first-class storage, and it stores up-to 128KB
of data. Once the full chunk is sent, it is flushed to database. After a while,
the data in Redis and database will be archived to [object storage](#uploading-traces-to-object-storage).
of data. Once the full chunk is sent, it is flushed a persistent store, either object storage(temporary directory) or database.
After a while, the data in Redis and a persitent store will be archived to [object storage](#uploading-traces-to-object-storage).
The data are stored in the following Redis namespace: `Gitlab::Redis::SharedState`.
......@@ -89,11 +89,11 @@ Here is the detailed data flow:
1. GitLab Runner picks a job from GitLab
1. GitLab Runner sends a piece of trace to GitLab
1. GitLab appends the data to Redis
1. Once the data in Redis reach 128KB, the data is flushed to the database.
1. Once the data in Redis reach 128KB, the data is flushed to a persistent store (object storage or the database).
1. The above steps are repeated until the job is finished.
1. Once the job is finished, GitLab schedules a Sidekiq worker to archive the trace.
1. The Sidekiq worker archives the trace to object storage and cleans up the trace
in Redis and the database.
in Redis and a persistent store (object storage or the database).
### Enabling live trace
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment