Replace all occurrences of
```foo
{% raw %}
bar
{% endraw %}
```
(which generates spurious empty lines in the rendered pre block) with
{% raw %}
```foo
bar
```
{% endraw %}
Also, fix some occurrences where the raw section is too large and
prevented interpretation of Jekyll directives.
This is the syntax used in the documentation of Jekyll itself:
https://raw.githubusercontent.com/jekyll/jekyll/master/docs/_docs/templates.md
FTR, done with two perl substitutions:
'^([\t ]*```[^\n]*
)([ \t]*\{% raw %\}[^\n]*
)' '$2$1'
and
'^([ \t]*\{% endraw %\}[^\n]*
)([\t ]*```[^\n]*
)' '$2$1'
and manually tweaks. A mechanical check would be most useful.
Signed-off-by: Akim Demaille <akim.demaille@docker.com>
Commands were failing because the DTR version number wasn't rendering. I fixed it on this page by replacing "dtr_version" with "page.dtr_version" in three places.
* dtr-rethinkdb now houses the rethinkcli which means we no longer need
to pull the rethinkcli container from hub.
Signed-off-by: Kyle Squizzato <kyle.squizzato@docker.com>
* HRM and Window Servers
Note:- Windows Server doesn't support Swarm Mode Routing Mesh yet (not considering Windows Server version 1709 which will be out soon). This means the hrm service is exposed only through the manager node's ip (since Windows Server cannot be a manager node). Therefore the DNS entry for the host name ('wordpress.example.org' in this example) should be mapped to the manager node and not to any of the Windows worker nodes.
* HRM with Windows Server
HRM with Windows Server
`-z` option for tar is used to filter through gzip. This should be excluded for the restore as it was not used during the backup.
The path is also fixed as it recreated the entire path within the volumes folder.
I think the actual documentation can be confusing.
In fact if we want to completely remove a DTR cluster, we can use the destroy command, it is fine.
But, if we just want to remove a replica from a cluster, it is better to use the remove command which is safer and cleaner.
The destroy command do not send a notification to the RethinkDB on the others replicas which can be dangerous and leads to unhealthy DTR clusters.
* Reword lots of instances of 'will'
* Reword lots of instances of won't
* Reword lots of instances of we'll
* Eradicate you'll
* Eradicate 'be able to' type of phrases
* Eradicate 'unable to' type of phrases
* Eradicate 'has / have to' type of phrases
* Eradicate 'note that' type of phrases
* Eradicate 'in order to' type of phrases
* Redirect to official Chef and Puppet docs
* Eradicate gratuitous 'please'
* Reduce use of e.g.
* Reduce use of i.e.
* Reduce use of N.B.
* Get rid of 'sexagesimal' and correct some errors
Added:
Note that, when planning for host storage, workflows based around `docker pull` through UCP will result in higher storage requirements on manager nodes, since `docker pull` through UCP results in the image being pulled on all nodes.
This has become an issue for some customers who did not plan to make additional storage available on UCP managers.
This request is based on an internal support case with Genesys ( A Docker Enterprise Customer ) who confirmed through our sales department ( who has relayed they've confirmed via engineering ) that we current support installing DTR on the same nodes as UCP manager nodes. Our documentation however says this is a requirement and Genesys has requested we update our documentation to reflect this which is the basis of this change. [https://docker.my.salesforce.com/5000f00001FBE2l](Docker Support Case 00033149)
* Added Language regarding FW Ports
Added additional language regarding system requirements and firewall ports to UCP installation.
* Update url to be relative instead of absolute
* Correct sync_schedule in ucp-description-file.md
The sync_schedule parameter in the auth.ldap section of the UCP Configuration file contains a seconds field, whilst the existing documentation claims this is omitted. Should explicitly call out inclusion of seconds field here, as this is usually omitted in a standard CRON entry.
* Update UCP config
Make explicit that the `sync_schedule` needs to include the seconds field, but it always needs to be zero.