I have created a role (commonrole) and applied to multiple nodes.
Now I want to override one of the attributes on 1 particular node to change to a different value.
So , created 1 more role (noderole) and applied that role after "commonrole "to this node but my node does not picks the new value (-Xmx2048m as mentioned below).
Sample common role-
{
"name": "commonrole",
"description": "Manages all nodes",
"run_list": [
"recipe[abc]"
],
"default_attributes": {
"catalina_opts": [
"-Dfile.encoding=UTF-8"
]
}
Sample noderole-
{
"name": "noderole",
"description": "Manages particular node",
"run_list": [
"role[commonrole]"
],
"default_attributes": {
"catalina_opts": [
"-Dfile.encoding=UTF-8",
"-Xmx2048m"
]
}
}
Am I missing something?
Arrays in node attributes are kind of weird. I've got a full write up on my site but basically this should result in the merged value being:
[
"-Dfile.encoding=UTF-8",
"-Dfile.encoding=UTF-8",
"-Xmx2048m"
]
or something similar. Also remember you won't see the attribute change immediately in the knife node show output, only after a successful converge.
Related
What I need to do is:
Create routing rules for each of my routes (http) assigned with just 2 frontends
Create one more routing rule, but with all available frontends assigned to it.
Currently I am using the copy like this
copy[
"name": "routingRules",
"count": mylengthvariable,
"input": {
"name": mynamearraywithsomeconcats,
"properties": {
"frontendEndpoints": the frontendpoint for this rule,
etc.. etcc
}
}
}
}
]
this is all well and good but I need to add yet another routing rule, but with more frontendpoints; which would like this: (this is placed outside of the copy above)
"name": "routingRules",
"input": {
"name": "extrarule",
"properties": {
"copy":[
{
"name" : "frontEndpoints",
"count": frontendpointcount,
"input" : {
"id": a list of frontendpoints
}
}
],
etc.. etc..
}
},
When I try this, I get the error because I am trying to add one more rule (I think)
I am seeking help on how to implement such a scenario.
Thanks in advance.
It is not possible to perform nested copies in ARM templates (unfortunately). There are some workarounds you can perform, such as expanding the sub-item in a variable, then referencing the variable.
Details and examples can be found here: https://learn.microsoft.com/en-us/azure/azure-resource-manager/templates/copy-resources
I would like to do a query matches against two properties of the same item in a sub-collection.
Example:
[
{
"name": "Person 1",
"contacts": [
{ "type": "email", "value": "person.1#xpto.org" },
{ "type": "phone", "value": "555-12345" },
]
}
]
I would like to be able to search by emails than contain xpto.org but,
doing something like the following doesn't work:
search.ismatchscoring('email','contacts/type,','full','all') and search.ismatchscoring('/.*xpto.org/','contacts/value,','full','all')
instead, it will consider the condition in the context of the main object and objects like the following will also match:
[
{
"name": "Person 1",
"contacts": [
{ "type": "email", "value": "555-12345" },
{ "type": "phone", "value": "person.1#xpto.org" },
]
}
]
Is there any way around this without having an additional field that concatenates type and value?
Just saw the official doc. At this moment, there's no support for correlated search:
This happens because each clause applies to all values of its field in
the entire document, so there's no concept of a "current sub-document
https://learn.microsoft.com/en-us/azure/search/search-howto-complex-data-types
and https://learn.microsoft.com/en-us/azure/search/search-query-understand-collection-filters
The solution I've implemented was creating different collections per contact type.
This way I'm able to search directly in, lets say, the email collection without the need for correlated search. It might not be the solution for all cases but it works well in this case.
In my ARM template I have a variable called "subnets" which can be of 3 types.
If it is of typeA then I want 4 subnets of the given names and addresses; if it's typeB then 2 subnets, and so on.
"variables": {
"subnets" : {
"typeA" : {
"network" : "3.0/24",
"directory" : "5.0/24",
"documents" : "8.0/24",
"security" : "10.0/24",
},
"typeB" : {
"directory" : "10.0/24",
"database" : "11.0/24",
},
"dmz" : {
"directory" : "12.0/24",
"database" : "15.0/24", }
}
}
In the ARM template I have a parameter which tells me what type to use. So I have a segment like the below which uses a condition to match on the subnetType being typeA and creates a virtual network accordingly.
{
"type": "Microsoft.Network/virtualNetworks",
"condition" : "[contains(parameters('subnetType'), 'typeA')]",
"apiVersion": "2018-10-01",
...
"copy" : [ {
"name" : "subnets",
"count" : "[length(array(variables('subnets').typeA))]",
"input": {
"name": "...",
"properties": {
"addressPrefix": "..."
}
}
} ]
}
}
As you can see above, I have a copy block within this VirtualNetwork resource, and I want to create the various subnets for the typeA network. I figure I could convert subnets.typeA to an array and get the length of it to loop over (that's the idea, I don't know if it actually works) but I am not clear how to extract the subnet name and addressprefix from my variable above.
so there are 2 issues here:
no way to loop object keys in arm templates
use of different resources in the template to create subnets
there is no way to work around the first limitation that I know of, whereas the second limitation is mostly due to you trying to work around the first one. I'd go for a completely different approach:
"networks": [
{
"name": "typeA",
"subnets": [
{
"name": "network",
"addressSpace": "3.0/24"
},
{
"name": "directory",
"addressSpace": "5.0/24"
},
{
"name": "documents",
"addressSpace": "8.0/24"
},
{
"name": "security",
"addressSpace": "10.0/24"
}
]
},
{
// second virtual network
},
{
// x virtual network
}
]
the main downside here - you'd have to have a nested deployment, because you cannot actually iterate array inside array, so you'd have to feed each object inside array into a deployment that would create a virtual network that can contain various subnets.
You can consult this link for an example of this exact approach or the official Azure Building Blocks thingie way of doing this (which is quite similar in the approach, but the implementation is different).
You could, get away with different resources instead of iterations, but that means you are less flexible and each time you make a change to the input everything breaks or just doesnt work like you think it would (your way of doing this would fall apart if dmz doesnt exist in that variable, you'll get a compilation error, similarly if you add another key to the object, say applicationgateway it will work, but that virtual network won't get created)
I am working on a skill, which asks the user for selection of an item from a particular list. Now user's most intuitive response will be to say something like first or second. But built in slots AMAZON.Number only does not capture input like first or second. Am I doing something wrong? Is it possible with some other hack?
Akshay,
AMAZON.Number is really only for numbers, where first, second and such are strings.
To do this, in the Developer console you would need to create a custom slot type called something like "positionChioces". Then enter in first, second, third, fourth, etc. Then add that slot type to your intent naming it choices, then picking "positionChioces".
This is from the JSON editor:
"types": [
{
"name": "positionChioces",
"values": [
{
"name": {
"value": "fourth"
}
},
{
"name": {
"value": "third"
}
},
{
"name": {
"value": "second"
}
},
{
"name": {
"value": "first"
}
}
]
From that point it would all depend on your code. You can pull that slot from your user then use some logic saying if response was first, it will be the fist item in your array and so on. Hope this helps.
D
Hi I am using CouchDB and assuming I have an articles document with the field users, containing an array of user IDs that are allowed to view this article.
Example scenario, there will be a paginated table view showing 10 articles per page, my controller will retrieve the first 10 articles from CouchDB then perform the access rights check one by one on the returned articles. But the current user may only have view access rights on say, 8 of them, therefore the table will only show 8 articles instead of 10.
What are the best practice of handling such situation besides implementing the access rights logic on the CouchDB layer?
To accomplish this, I would simply use a view keyed on the users field:
function (doc) {
doc.users.forEach(function (user) {
emit([ user ]);
});
}
I emitted an array with just 1 item in this case. I presume you'd also emit something like doc.created in order to have your articles sorted, you would simply add them after user in that array.
The view results would look something like:
{
"rows": [
{ "id": "<article-1>", "key": [ "<user-1>", "<created>" ] },
{ "id": "<article-2>", "key": [ "<user-1>", "<created>" ] },
{ "id": "<article-3>", "key": [ "<user-1>", "<created>" ] },
{ "id": "<article-1>", "key": [ "<user-2>", "<created>" ] },
{ "id": "<article-1>", "key": [ "<user-3>", "<created>" ] }
]
}
You can simply paginate like you normally would with CouchDB. You simply use start_key=["<user-1>"]&end_key=["<user-1>","\ufff0"] in addition to the usual paging limit=10&skip=0 for page 1, limit=10&skip=10 for page 2, etc.