Lowest Common Manager


class OrgChart:

    def __init__(self, name, manager):
        self.name = name
        self.manager = manager
        self.directReports = []

    def __repr__(self):
        return f'{self.name}'

def getLowestCommonManager(topManager, reportOne, reportTwo):
    report_one_manager = reportOne.manager
    report_one_chain = list()
    while report_one_manager:
        report_one_manager = report_one_manager.manager

    report_two_manager = reportTwo.manager
    report_two_chain = list()
    while report_two_manager:
        report_two_manager = report_two_manager.manager

    common_manager = None
    for a, b in zip(report_one_chain, report_two_chain):
        if a == b:
            common_manager = a

    return common_manager

if __name__ == '__main__':
    org = [
        {"directReports": ["B", "C"], "id": "A", "name": "A"},
        {"directReports": ["D", "E"], "id": "B", "name": "B"},
        {"directReports": ["F", "G"], "id": "C", "name": "C"},
        {"directReports": ["H", "I"], "id": "D", "name": "D"},
        {"directReports": [], "id": "E", "name": "E"},
        {"directReports": [], "id": "F", "name": "F"},
        {"directReports": [], "id": "G", "name": "G"},
        {"directReports": [], "id": "H", "name": "H"},
        {"directReports": [], "id": "I", "name": "I"}

    top_manager = None
    members = {}
    for member in org:
        name = member['name']
        if name not in members:
            member_obj = OrgChart(name, None)
            top_manager = member_obj
            members[name] = member_obj

        member_obj = members[name]

        for direct_report in member['directReports']:

            if not direct_report in members:
                direct_report_obj = OrgChart(direct_report, member_obj)
                direct_report_obj.name = direct_report
                members[direct_report] = direct_report_obj
                direct_report_obj = members[direct_report]


    print(getLowestCommonManager(top_manager, members['F'], members['H']))

Longest Peak

package com.vishpat.practice.ds.array;

enum States {

public class LongestPeak {
    public static int longestPeak(int[] array) {
        States currentState = States.UNKNOWN;
        States prevState = States.UNKNOWN;
        int longestSize = 0;
        int ascendingStartIdx = -1;
        int descendingEndIdx = -1;

        if (array.length < 3) {
            return 0;

        for (int i = 0; i < array.length - 1; i++) {
            prevState = currentState;

            if (array[i + 1] > array[i]) {
                switch (prevState) {
                    case UNKNOWN:
                        currentState = States.ASCENDING;
                        ascendingStartIdx = i;
                    case ASCENDING:
                        currentState = States.ASCENDING;
                    case DESCENDING:
                        currentState = States.ASCENDING;
                        ascendingStartIdx = i;

            if (array[i + 1] < array[i]) {
                switch (prevState) {
                    case UNKNOWN:
                        currentState = States.DESCENDING;
                    case ASCENDING:
                        currentState = States.DESCENDING;
                        descendingEndIdx = i + 1;
                    case DESCENDING:
                        currentState = States.DESCENDING;
                        descendingEndIdx = i + 1;

            if (ascendingStartIdx >= 0 && descendingEndIdx >= 0 &&
                ascendingStartIdx < descendingEndIdx) {
                int size = descendingEndIdx - ascendingStartIdx + 1;
                if (size > longestSize && size > 2) {
                    longestSize = size;
        return longestSize;

Max 2 integers from an array

def sol2(data):
    max_idx1 = 0
    max_idx2 = 1
    max_sum = data[0] + data[1]

    for next_idx in range(2, len(data)):
        if data[max_idx1] + data[next_idx] > max_sum and max_idx1 != next_idx:
            max_sum = data[max_idx1] + data[next_idx]
            max_idx2 = next_idx

        if data[max_idx2] + data[next_idx] > max_sum and max_idx2 != next_idx:
            max_sum = data[max_idx2] + data[next_idx]
            max_idx1 = next_idx

    print(max_idx1, max_idx2)

if __name__ == "__main__":
    data = [1, 5, 3, 2, 4]

Simple in-memory file system

class Dir:

    def __init__(self, parent, name):
        self.parent = parent
        self.name = name
        self.children = dict()

    def mkdir_path(self, path):
        sub_dirs = path.split('/')
        current_dir = self
        for child_dir_name in sub_dirs:
            current_dir = current_dir.mkdir(child_dir_name)

    def mkdir(self, name):
        dir = Dir(self, name)
        self.children[name] = dir
        return dir

    def get_child_dir(self, name):
        return self.children[name]

    def get_child_dir_path(self, path):
        sub_dirs = path.split('/')
        current_dir = self
        for sub_dir in sub_dirs:
            current_dir = current_dir.children[sub_dir]
        return current_dir

    def write_file(self, name, contents):
        self.children[name] = contents

    def write_file_path(self, path, contents):
        sub_dirs = path.split('/')
        file_name = sub_dirs[-1]
        current_dir = self
        for sub_dir in sub_dirs[:-1]:
            current_dir = current_dir.children[sub_dir]
        current_dir.write_file(file_name, contents)

    def read_file(self, name):
        return self.children[name]

    def read_file_path(self, path):
        sub_dirs = path.split('/')
        file_name = sub_dirs[-1]
        current_dir = self
        for sub_dir in sub_dirs[:-1]:
            current_dir = current_dir.children[sub_dir]
        return current_dir.read_file(file_name)

    def __repr__(self):
        return f'{self.name} {self.children}'

if __name__ == "__main__":
    fs = Dir(None, '/')
    fs.get_child_dir('tmp').get_child_dir('1').write_file('test', 'data')
    contents = fs.get_child_dir('tmp').get_child_dir('1').read_file('test')
    assert contents == 'data', "contents of the file don't match"

    fs.write_file_path('tmp/1/2/3/4/4.dat', '4')
    contents = fs.read_file_path('tmp/1/2/3/4/4.dat')
    assert contents == '4', 'Correct content not found'

Check if concatenation of any permutation of given list of arrays generates the given array

Solution for geeksforgeeks problem

def solve(arr, sub_arrs):

    sub_arr_membership = {}
    curr_indx = [-1 for i in range(0, len(sub_arrs))]
    for idx, sub_arr in enumerate(sub_arrs):
        for sub_arr_idx, element in enumerate(sub_arr):
            sub_arr_membership[element] = (idx, sub_arr_idx)

    for element in arr:
        idx, sub_arr_idx = sub_arr_membership[element]
        if curr_indx[idx] > sub_arr_idx:
            curr_indx[idx] = sub_arr_idx


Custom kubernetes ingress authentication using Hashicorp Vault

This article explains how to implement an ngnix ingress authentication (basic http auth) mechanism for securing your kubernetes (k8s) application endpoints using the credentials stored in hashicorp vault. This can be achieved easily by writing a simple http auth (python-flask) app that talks with the vault. The workflow for custom authentication is explained well in this article.

The basic idea to is modify the opensource httpbin app by adding a new endpoint such as vault-auth. The code this endpoint is going to basic-auth except the http auth credentials will be validate against hashicorp vault instead of the passed parameters. The modified code can then be deployed as a container and as an auth service in your kubernetes cluster.

All the applications whose ingress endpoints need to be secured have to add the following line to their ingress deployment yaml

nginx.ingress.kubernetes.io/auth-url: http://auth.default.svc.cluster.local/vault-auth/x/y

Where auth.default.svc.cluster.local is the service that is running the modified httpbin app

Back to Storage at Cisco

I joined the Webex Storage IAAS (infrastructure as a service) group three weeks ago after a stint of three and half years at Nuage. I had a great experience at Nuage, especially during my first two years during which I got to work quite a bit on Openstack, Vmware cloud, Kubernetes and Openshift. I was super fortunate to make lots of opensource contributions during this period and I think that really helped in getting my new job at Cisco.

At Cisco I am part of the Storage team and really excited to back in the storage domain after learning a lot about networking in last four and half years.


I recently started working on a big C project at Nuage. I have worked in C at Netapp for 9 years and was looking forward to this new work as result. Man, I am so glad so be able to program in C again. I was also happy to realize that my C chops are still quite good.

However having worked with Java/Python for past 2 – 3 years and being used to the awesome Intellij IDE, I really wasn’t enjoying working with vim again. My complaints about vim were two fold. First writing code in C is quite verbose, especially for the low level stuff I was working on. (I really missed Java’s awesome collection library and stream API. Those things made writing software such a joy). The second complaint about vim is lack of auto completion, which makes the first problem all the more worse.

I tried overcoming this problem by using the YouCompleteMe plugin for vim but it wasn’t working quite well with my C project. (C projects do not have a standard way to compile and build, each one of them has a different way of building things). Initially the auto-completion with YouCompleteMe really sucked, until I discovered the bear tool. The bear tool lets you generate a compilation database for your project (You just prepend your make command with bear) and once this is done, autocompletion worked as charm in Vim. Morever YouCompleteMe dynamically complies the C file so, I get syntax errors while writing the code itself. What a timesaver and productivity booster.

Opensource contributions

I always wanted to contribute to some major Opensource project since my engineering student days (more that 10 years ago). In those days I was looking to contribute the Linux kernel project but never did I get a chance.

Howpever as part of my work for Nuage, I have been (finally) able to contribute to Opensource projects. Specifically I have contributed to the following Opensource projects

  • Openstack Puppet Neutron
  • Openshift Ansible
  • Openstack Fuel Plugin for Nuage