Patrick,

Thanks for your review.

Regarding the severity leaf: multiple severities are already covered when 
severity is specified because the default severity comparison is all messages 
of the specified severity and greater. You can use of the optional 
pattern-match string to select a message with a specific priority (includes 
severity and facility). If you need further filtering capability, you could add 
this through augmentation of the log-selector container.

Regarding the specification for the remote destination: in an earlier review it 
was required that the model conform to the recommendation in Section 3 of 
draft-schoenw-netmod-yang-pattern-00. Your proposal does not follow that 
pattern.

Regarding the leaf nodes to be added request:

- The presence or absence of a leaf in the log-actions container indicates 
whether the intended action is enabled/disabled. This pattern is used for all 
log actions.
- The proposed ietf-syslog model does not include operational state or 
notifications. Operational state in the model's config section would have to be 
flagged as such. Please refer to the "State Data" statement from RFC 6020 
section 4.2.3.
- A custom-prefix leaf to prepend a custom signature for each log message could 
be added though augmentation. If there is a consensus that this should be added 
for all implementations I will gladly add it.

Thanks,

Clyde

From: "Hinshaw, Patrick" <[email protected]<mailto:[email protected]>>
Date: Tuesday, March 1, 2016 at 11:22 PM
To: 
"[email protected]<mailto:[email protected]>"
 
<[email protected]<mailto:[email protected]>>
Subject: draft-ietf-netmod-syslog-model
Resent-From: 
<[email protected]<mailto:[email protected]>>
Resent-To: 
<[email protected]<mailto:[email protected]>>
Resent-Date: Tuesday, March 1, 2016 at 11:54 PM

Hello,

There are a few leaves in this current draft that I believe should be altered, 
as well as a few that should be added.

First, the severity leaf node in the syslog-severity grouping could be changed 
to a leaf-list node. This will enable a user to choose any combination of 
severities, instead of being strictly limited to a range. This functionality 
would be used when the severity-operator leaf is set to “equal”.

Second, in the remote container, there is a list called “destination” with the 
key “name” of type string. Another way to do this would be to change this type 
to inet:host, as well as adding a new leaf node that holds the resolved IP 
address of type inet:ip-address (read-only). We would in turn delete the leaf 
node “address” under both the tcp and udp choice statement. This would 
eliminate the need to create an arbitrary name for each server, having the 
hostname as the list key seems to be the more intuitive approach.

Lastly, there are a few leaf nodes that I  believe should added to the remote 
container:
1)      An administrative state leaf node to enable/disable the remote server.
2)      An operational state leaf node to see the operational state of the 
remote server (This is read-only).
a.       Both of these nodes would allow the user to be aware of the status of 
their syslog servers, something this model is currently lacking.
b.      Both would be enumerated types with values enable and disable.
3)      A custom-prefix leaf node of type string allowing a custom signature to 
be prepended to each syslog message.

Here’s a tree of the remote container made in pyang showing the changes I’ve 
proposed. Note that these changes will affect other containers using the 
severity node from the syslog-severity grouping:

module: draft-ietf-netmod-syslog-model-06
….
         +--rw remote
         |  +--rw destination* [name]
         |     +--rw name                    inet:host
         |     +--ro resolved-ip?            inet:ip-address
         |     +--rw (transport)
         |     |  +--:(tcp)
         |     |  |  +--rw tcp
         |     |  |     +--rw port?      inet:port-number
         |     |  +--:(udp)
         |     |     +--rw udp
         |     |        +--rw port?      inet:port-number
         |     +--rw log-selector
         |     |  +--rw (selector-facility)
         |     |  |  +--:(no-log-facility)
         |     |  |  |  +--rw no-facilities?   empty
         |     |  |  +--:(log-facility)
         |     |  |     +--rw log-facility* [facility]
         |     |  |        +--rw facility             union
         |     |  |        +--rw severity*            union // now a leaf-list 
node
         |     |  |        +--rw severity-operator?   enumeration 
{selector-severity-operator-config}?
         |     |  +--rw pattern-match?   string 
{selector-match-processing-config}?
         |     +--rw destination-facility?   identityref
         |     +--rw source-interface?       if:interface-ref
         |     +--rw custom-prefix?          string
         |     +--rw administrative-state?   enumeration
         |     +--ro operational-state?      enumeration
         |     +--rw syslog-sign! {signed-messages-config}?
         |        +--rw cert-initial-repeat    uint16
         |        +--rw cert-resend-delay      uint16
         |        +--rw cert-resend-count      uint16
         |        +--rw sig-max-delay          uint16
         |        +--rw sig-number-resends     uint16
         |        +--rw sig-resend-delay       uint16
         |        +--rw sig-resend-count       uint16
….

Please let me know what your thoughts are on these proposed changes.

Respectfully,

Patrick Hinshaw
[email protected]<mailto:[email protected]>

_______________________________________________
netmod mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/netmod

Reply via email to