Skip to content

documenting rateLimit wich leads to the confusion of #28641#670

Merged
mdlinville merged 2 commits intodocker:masterfrom
nilsotto:feature/journald_rate
Dec 19, 2016
Merged

documenting rateLimit wich leads to the confusion of #28641#670
mdlinville merged 2 commits intodocker:masterfrom
nilsotto:feature/journald_rate

Conversation

@nilsotto
Copy link
Contributor

Describe the proposed changes

Documenting the fact that journald by default will skip messages if it exceeds a certain rate

Related issue

Relates to my issue #28641 where i blame the logging driver, when the problem were system configuration.

@johndmulhausen
Copy link
Contributor

johndmulhausen commented Nov 21, 2016

mdlinville pushed a commit to wallyqs/docker.github.io that referenced this pull request Nov 22, 2016
The format for node advertisement in swarm and docker engine (for
networking) is the same, and can be shared. Doing so allows us to start
toward removing the swarm-join container, and improves our HA situation
after engine discovery is run. Resolves docker#670.

Specifically, swarm was previously somewhat HA: if you set up 1
controller, 2 replicas and 1 node, then you could lose the controller
and still remain functional unless the swarm node needed to restart.
This change improves HA by removing the above failure mode as long as
engine discovery has been run and is up to date.

See also docker-archive/classicswarm#1368 which implemented this flag, and lays out the
future scenario we hope to achieve, and also docker#1032 which is our goal of
removing swarm join with automatical engine discovery.
mdlinville pushed a commit to amincheloh/docker.github.io that referenced this pull request Nov 28, 2016
The format for node advertisement in swarm and docker engine (for
networking) is the same, and can be shared. Doing so allows us to start
toward removing the swarm-join container, and improves our HA situation
after engine discovery is run. Resolves docker#670.

Specifically, swarm was previously somewhat HA: if you set up 1
controller, 2 replicas and 1 node, then you could lose the controller
and still remain functional unless the swarm node needed to restart.
This change improves HA by removing the above failure mode as long as
engine discovery has been run and is up to date.

See also docker-archive/classicswarm#1368 which implemented this flag, and lays out the
future scenario we hope to achieve, and also docker#1032 which is our goal of
removing swarm join with automatical engine discovery.
Copy link

@mdlinville mdlinville left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for tackling this. Please consider my rewording suggestions and please rebase against master and resolve the merge conflict if possible. Let me know if you need help with the rebase.

## Journald configuration

Docker hosts with many containers may produce massive amounts of log-messages.
The default configuration of journald is to limit the number of messages stored per service per time-unit.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please put back-ticks around journald. Consider rewriting to something like "By default, journald limits the number of messages stored per service per time-unit."

Docker hosts with many containers may produce massive amounts of log-messages.
The default configuration of journald is to limit the number of messages stored per service per time-unit.

Applications with need for large-scale logging wil probably need to configure `RateLimitIntervalSec` and `RateLimitBurst` in the journald configuration file.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

typo: "wil"

Consider rewording because applications don't configure. "If your applications need large-scale logging, configure RateLimitIntervalSec and RateLimitBurst in the journald configuration file."

@mdlinville mdlinville force-pushed the feature/journald_rate branch from 6bda312 to 0ddda59 Compare December 19, 2016 18:42
@mdlinville
Copy link

I rebased your commits and used this as an opportunity to improve the formatting in the rest of the file. I'll merge after the PR finishes running tests.

@mdlinville mdlinville merged commit 4e2ee1a into docker:master Dec 19, 2016
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants