Today we continue discussing metrics and their collection from services.
We observed that one way to keep metrics true is to collect them directly from the service where the service makes a call to push the metric.
One example of this is that we can count the number of success from the code by calling
StatsdClient.Metrics.Counter("myapi.post.Success");
Statsd is a client for Graphite-Graphana stack and makes it easy to push the metrics to Graphite so that they show in the charts in Graphana.
Statsd exposes the notion of timers as well.  A timer captures the duration of an operation and we can use it in our constructor and destructor to measure the lifetime of an operation and this can be used with the metrics.
Statsd batches up the metrics it pushes to Graphite. Therefore the timing window for the statsd flush interval must be at least as long as the graphite's minimum time otherwise Graphite will drop packets.
 And we must also be careful to turn legacyNamespace off in Graphite otherwise the statsd measurements will show up in two or three different folders.
Statsd is not the only client. There are statsd, collectd, bucky and logster. Statsd is the original client and the most used of them all. Collectd is written in EPEL in older versions. Bucky connects them both because it listens to both the packets from statsd and collectd and sends them to Graphite. Logster can tail your log file and send metrics back to Graphite.
Some metrics can be confusing. For example, count and rate are two independent metrics. Count measures the total number of times an event happened.  Rate measures how many times the event happened per unit of time.
Metrics can collect over time. There are retention rules mentioned which can dictate how long the metrics need to be held in the database files. Moreover, these files are fixed size.

Convert a BST into inorder singly linked list in place.
Node toList(node root)
{
If (root == null) return root;
If (root.left != null)
{
var left = toList(root.left);
for( ; left.right != null; left = left.right);
Left.right = root;
}
If (root.right != null)
{
var right = toList(root);
for(; right.left != null; right = right.left);
root.right = right;
}
Return root;
}

Convert a BST into preorder singly linked list in place.
Node toPreList(node root)
{
If (root == null) return root;
var temp = root.right;
var left = null;
If (root.left != null)
{
left = toPreList(root.left);
root.right = left;
for( ; left != null && left.right != null; left = left.right);
//Left.right = temp;
}
If (temp != null)
{
var right = toPreList(temp);
left.right = right;
}
Return root;
}

Convert a BST into postorder singly linked list in place.
Node toPostList(node root)
{
If (root == null) return root;
var temp = null;
var left = null;
var right = null;
If (root.left != null)
{
left = toPostList(root.left);
temp = left;
for( ; left != null && left.right != null; left = left.right);
}
If (root.right != null)
{
var right = toPostList(temp);
if (temp == null) temp = right;
if (left != null) left.right = right;
for( ; right != null && right.right != null; right = right.right);
}
if (temp == null) temp = right;
if (right != null) right.right = root;
if (temp == null) temp = root;
return temp;
}

In the above methods, we have used singly linked list. We can also make doubly linked list by adding links in the reverse direction
void ToDLL(Node head, Node prev)
{
if (head == null) return;
head.prev = prev;
var prev = head;
ToDLL(head.next, prev);
}
Test cases:
NULL,
1,
1->2,
1->2->3