Rico's Blog

The Hidden Danger of select on a Shared ioredis Connnection

Sun, 15 Mar 2026 03:37:00 GMT

A subtle but painful bug recently cost me a significant amount of debugging time. This post documents what went wrong, what the root cause was, and how to avoid the same trap.

The Setup

I have a Node.js application using a single persistent ioredis client shared across API endpoints. Two of those endpoints read from different Redis databases:

/endpoint0 reads from db 0 using a straightforward GET:

const res = await client.get(someKey)

/endpoint2 reads from db 2 using a pipeline with SELECT:

const [res1, res2] = await client.multi().select(2).hget(hash, key).exec()

/endpoint0 was added well after /endpoint2 was already in production. Shortly after it was introduced, an intermittent failure started appearing — calls to /endpoint0 would begin returning incorrect results some time after a fresh deployment, and a redeployment would temporarily resolve it.

The Investigation

The bug manifested across all environments — dev, UAT, and production — which ruled out environment-specific configuration issues. I checked dependency declaration order, environment variable handling, and the deployment pipeline itself. None of those investigations turned up anything.

The breakthrough came when I inspected the output of CLIENT LIST directly on the Redis server:

id=501766576 addr=192.168.1.57:61819 fd=36 name= age=701 idle=623 flags=N db=0 sub=0 psub=0 multi=-1 qbuf=0 qbuf-free=0 obl=0 oll=0 omem=0 events=r cmd=hget

id=501692280 addr=192.168.1.57:24319 fd=10 name= age=3619 idle=4 flags=N db=2 sub=0 psub=0 multi=-1 qbuf=0 qbuf-free=0 obl=0 oll=0 omem=0 events=r cmd=get

The db=2 field on the second connection immediately stood out. A connection that should have been operating on db 0 was sitting on db 2 — which pointed directly at the select call in /endpoint2.

The Root Cause

SELECT in Redis — and by extension client.select() or client.multi().select() in ioredis — is a stateful, connection-level operation. It does not scope to a single command or pipeline; it changes the active database for that connection and the change persists until another SELECT is issued.

When a shared connection is used, any call to SELECT affects every subsequent command on that connection, regardless of which endpoint or function issued them. In this case, once /endpoint2 ran and switched the connection to db 2, any subsequent call from /endpoint0 was unknowingly querying db 2 instead of db 0. Refer to this diagram for a better understanding of the problem:

The Fix

In this specific case, the simplest resolution was to migrate the data for /endpoint2 into db 0, eliminating the need for SELECT entirely:

const res = await client.hget(hash, key)

The more robust and general solution — especially in applications where multiple databases are genuinely required — is to create a dedicated ioredis client per database. Calling SELECT at runtime on a shared connection is a latent bug waiting to surface, and it will do so in ways that are difficult to reproduce and trace.

Takeaway

If you are using ioredis with multiple Redis databases in a shared-connection setup, avoid SELECT at runtime. Provision one client per database instead, and make the database assignment explicit at initialization. It is a small upfront cost that eliminates an entire class of hard-to-debug, timing-dependent failures.

How I upgraded my home lab from Nginx to Caddy

Sun, 15 Mar 2026 03:37:00 GMT

I have a registered domain with Cloudflare — the blog you're reading right now is just one of the sites running on it. I also host a password manager, Vaultwarden, behind a WAF.

When I first set up my home lab, I used Nginx as the middleman between Cloudflare and my applications, with Let's Encrypt handling SSL. For extra protection, I ran a Redis server to dynamically manage a whitelist of allowed IPs — only requests from those IPs could reach protected sites like Vaultwarden or the Nextcloud instance I share with my family. I built a SvelteKit app (publicly accessible) to manage the whitelist, with authentication handled by Pocketbase. The actual WAF check happened inside Nginx via the Redis Lua module.

It worked, but it had a few pain points:

Every single request to a protected site triggered an IP whitelist lookup^footnote
Caching helped with performance, but introduced a lag — a newly whitelisted IP could take up to 10 seconds to be recognized
Wildcard SSL certificate from Let's Encrypt needs manual renewal every 3 months

Here's the architecture diagram of that original setup:

Then I came across Caddy. I read a post from someone who'd migrated their entire stack from Nginx to Caddy, got curious, and started experimenting a few weeks ago. I followed this Youtube tutorial^tutorial and was honestly blown away — automatic SSL certificate renewal alone made it worth the switch.

I also found a much cleaner approach to IP whitelisting. Instead of checking IPs inside my home lab on every request, I moved that logic to Cloudflare, managing the whitelist via their [list items API]:

PUT https://api.cloudflare.com/client/v4/accounts/{account_id}/rules/lists/{list_id}/items

A local SQLite database (used by the better-auth library) keeps track of the listed IPs on my end. The big win here: disallowed requests to the protected apps never even reach my home router.

Here's what the new stack looks like:

The original Excalidraw file for both diagrams is here.

References

How to wakeup/shutdown a PBS server from proxmox

Wed, 11 Mar 2026 22:21:00 GMT

Intro

I have the following simple backup setup for my Proxmox server. Since backup schedule is weekly, it would make a lot of sense to bring the backup server up only when it's needed. I decided to make my own script after seeing one on ^reference Proxmox forum.

Setup

Here are the steps

configure the PBS server's BIOS so Wake-on-lan feature is enabled
install wakeonlan program on the Proxmox server if it's not already installed

apt install wakeonlan

set up password-less login from Proxmox server

ssh-keygen -t ed25519
ssh-copy-id root@$PBS

create /usr/local/bin/vzdump-hook-script as follow and make it executable

#!/bin/bash
IP=<LAN_IP_OF_PBS>
MAC=<MAC_ADDRESS_OF_PBS>
# wait for at least  $COUNT pings
COUNT=80

# check if wakeonlan is installed
which -s wakeonlan
if [ $? -ne 0 ]; then
    echo "Please install wakeonlan first"
    exit 1
fi
if [ "$1" == "job-init" ]; then
    if ! ping -c 1 -W 1 "$IP" >/dev/null 2>&1; then
        /usr/bin/wakeonlan $MAC   #(mac Address of PBS Server)
        ping $IP -c $COUNT
    fi
fi

if [ "$1" == "job-end" ]; then
    ssh root@$IP "/usr/sbin/poweroff < /dev/null &"
fi

exit 0

update /etc/pve/jobs.cfg to add this line

script /usr/local/bin/vzdump-hook-script

Reference

How to handle file uploads using actix-web

Tue, 09 Aug 2022 23:21:59 GMT

In this tutorial I'll demonstrate how to handle upload with additional data fields using one of the most popular Rust web frameworks - actix-web, which has become my go-to web framework when developing in Rust.

We'll start by creating a binary Rust package

cargo new doc-demo

Then under the project root, run

cargo add actix-web actix-multipart

With your favorite editor, open src/main.rs and copy/paste the following code

use serde::Serialize;
use actix_multipart::Multipart;
use futures_util::TryStreamExt as _;

use actix_web::{ post, App, Error as ActixError, HttpResponse, HttpServer };

#[derive(Serialize)]
struct Stats {
    lines: usize,
    #[serde(skip_serializing_if = "Option::is_none")]
    word_count: Option<usize>
}

#[post("/upload_stats")]
async fn upload_stats(
    mut payload: Multipart,
) -> Result<HttpResponse, ActixError> {
    let mut file_data = Vec::<u8>::new();
    let mut layout: Option<String> = Some("simple".to_owned());
    while let Some(mut field) = payload.try_next().await? {
        let content_disposition = field.content_disposition();
        let field_name = content_disposition.get_name().unwrap();
        match field_name {
            "file" => {
                while let Some(chunk) = field.try_next().await? {
                    file_data.extend_from_slice(&chunk);
                }
            }
            "layout" => {
                let bytes = field.try_next().await?;
                layout = String::from_utf8(bytes.unwrap().to_vec()).ok();
            }
            _ => {}
        }
    }
    let file_content = std::str::from_utf8(&file_data)?;
    let mut i = 0;
    let mut word_count=0;
    for line in file_content.lines() {
        word_count+=line.chars().count();
        i += 1;
    }
    let word_count_res = if layout.unwrap() == String::from("advanced") {
        Some(word_count)
    } else {
        None
    };
    Ok(HttpResponse::Ok().json(Stats {
        lines: i,
        word_count: word_count_res
    }))
}

#[actix_web::main]
async fn main() -> std::io::Result<()> {
     HttpServer::new(move || {
        App::new()
            .service(upload_stats)
    })
    .bind(("127.0.0.1", 8080))?
    .run()
    .await
}

This simple web application has only one single POST endpoint that will accept

a field named file that points to a file in client's file system
an optional field named layout, its value is defaulted to simple

By default the output will be the line count of the file being uploaded, but a characters result that represents the number of characters in the file will be added if layout is set to advanced. So for example,

curl http://localhost:8080/upload_stats -X POST -F 'file=@Cargo.toml'

returns something like

{"lines": 13}

While

curl http://localhost:8080/upload_stats -X POST -F 'file=@Cargo.toml' -F 'layout=advanced'

might produce something like

{"lines": 13, "characters": 311}

p.s. here's the full content of Cargo.toml

[package]
name = "doc-demo"
version = "0.1.0"
edition = "2021"

# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

[dependencies]
actix-multipart = "0.4.0"
actix-web = "4.1.0"
futures-util = "0.3.21"
serde = { version = "1.0.136", features = ["derive"] }
serde_json = "1.0.81"

Add error code(s) to Hapijs output

Sat, 08 Jun 2019 15:51:01 GMT

Joi validation is powerful and easy to work with, however it's not always obvious or easy to add stuff like error code(s) to the hapijs response. This post will show you a way (or two) to deal with that problem.

Step 1, assign error codes to each validation error

Quick example:

server.route({
  method: 'POST',
  path: '/person',
  options: {
    validate: {
      payload: {
        firstName: Joi.string()
          .min(5)
          .max(10)
          .required()
          .error(errors => {
            errors.forEach(err => {
              switch (err.type) {
                case 'any.empty':
                case 'any.required':
                  err.message = 'Firstname should not be empty!'
                  err.context = {
                    errorCode: 111
                  }
                  break
                case 'string.min':
                  err.message = `Firstname should have at least ${
                    err.context.limit
                  } characters!`
                  err.context = {
                    errorCode: 121
                  }
                  break
                case 'string.max':
                  err.message = `Firstname should have at most ${
                    err.context.limit
                  } characters!`
                  err.context = {
                    errorCode: 131
                  }
                  break
                default:
                  break
              }
            })
            return errors
          })
      }
    }
  },
  handler: async request => {
    // todo: handle saving the paylod
    console.log('to save', request.payload)
    return { result: 'ok' }
  }
})

Step 2, customize failAction when creating Hapi server

const Boom = require('@hapi/boom')
const server = Hapi.Server({
  // ...
  routes: {
    validate: {
      failAction: (request, h, err) => {
        const firstError = err.details[0]
        if (firstError.context.errorCode !== undefined) {
          throw Boom.badRequest(err.message, {
            errorCode: firstError.context.errorCode
          })
        } else {
          throw Boom.badRequest(err.message)
        }
      }
    }
  }
})

Step 3, customize response

The reason this step is needed is because Hapi would strip out the injected errorCode attribute created by step 2

server.ext('onPreResponse', (request, h) => {
  const response = request.response
  if (!response.isBoom) {
    return h.continue
  }
  const { data } = response
  if (data !== undefined) {
    response.output.payload = {
      ...response.output.payload,
      ...data
    }
  }
  return h.continue
})

With the above setup, request

curl http://localhost:3000/person -d ''

would result in

{"statusCode":400,"error":"Bad Request","message":"child \"firstName\" fails because [Firstname should not be empty!]","errorCode":111}

By default abortEarly: true is set Hapi, if multiple error codes are desired, only Step 2 needs to be adjusted to

const server = Hapi.Server({
  // ...
  routes: {
    validate: {
      options: {
        abortEarly: false
      },
      failAction: (request, h, err) => {
        const errorCodes = err.details
          .map(e => e.context.errorCode)
          .filter(e => e !== undefined)
        throw Boom.badRequest(err.message, { errorCodes })
      }
    }
  }
})

request

curl http://localhost:3000/person -d 'firstName='

would return

{"statusCode":400,"error":"Bad Request","message":"child \"firstName\" fails because [Firstname should not be empty!, Firstname should have at least 5 characters!]","errorCodes":[111,121]}

If you need to add error code in your application code, you can simply achieve that by returning a Boom like the following

return Boom.badRequest('Your error message here', { errorCode: YOUR_CODE })

I've composed a gist in case you want to save some typings in trying out the code. Cheers!

I just built my first standing desk

Sun, 23 Mar 2014 21:38:00 GMT

So it's for real

I've always wanted to build a standing desk using Ikea parts. Other people have done it with Finnvard height adjustable table legs (http://www.ikea.com/us/en/catalog/products/00144763/) and a table top. My plan was put off because the Long Island Ikdea store discontinued the Finnvard legs for some time. I was lucky to find them in-store again this weekend and Voilà - my dream desk is finally built:

I bought this $40 stool just incase I need to take a short break from standing: http://www.ikea.com/us/en/catalog/products/50199215/

I also like the fact that I am able to stuff my laser printer onto the shell: {% img center /images/standing_desk_4.JPG %}

Here's the recipe

1 x Linnmon table top, http://www.ikea.com/us/en/catalog/products/50251350/, $40

2 x Finnvard table legs http://www.ikea.com/us/en/catalog/products/00144763/, $30 ea

1 x Tertial Work lamp, http://www.ikea.com/us/en/catalog/products/20370383/, $9

Total: $109+Tax

Note

I didn't do some serious hack to increase the height of the table legs like this guy did http://www.ikeahackers.net/2014/02/convert-the-finnvard-into-a-height-adjustable-standing-desk.html because the legs are able to reach upto 36 5/8, with 1" for the table top, the final result is 37 5/8, which is pretty close to my ideal desk height (the height where elbows can rest on the table top).

Install Gitlab(6.4) on Raspberry PI

Sat, 11 Jan 2014 14:31:00 GMT

I am a big fan of both Raspberry PI and Gitlab so it kinda bugs me when my attempts to install Gitlab onto RPI didn't succeed because of failure to install therubyracer gem. Others experienced the similar problems I encountered: http://www.raspberrypi.org/phpBB3/viewtopic.php?t=32716&p=397934, by following most of user dpenezic's instruction I finally made it to install Gitlab (currently at version 6.4) onto my RPI (512MB Ram but only 384MB is available to the system as I allocate the rest to GPU). So here are what I did:

Steps

Follow https://github.com/gitlabhq/gitlabhq/blob/6-4-stable/doc/install/installation.md until "Install Gems"
Install libv8 (https://github.com/cowboyd/libv8)

[update: added git-svn to the list on 1/17/2014]

sudo apt-get install -y subversion git-svn [ -d ~/tmp ] || mkdir ~/tmp cd ~/tmp git clone https://github.com/cowboyd/libv8 cd libv8 bundle install

be patient, the following command takes a while

bundle exec rake clean build binary sudo gem install pkg/libv8-3.11.8.17-armv6l-linux.gem
Modify /home/git/gitlab/Gemfile (and .lock) to skip installation of libv8 (as it's installed through the above step) and the rubyracer

cd /home/git/gitlab sudo -u git -H editor Gemfile # and remove the line: gem "therubyracer" sudo -u git -H editor Gemfile.lock # and removed the following lines libv8 (3.16.14.3) therubyracer (0.12.0) libv8 (~> 3.16.14.0) ref
Install node.js, you can take a look at the script I came up with to compile node.js in RPI: https://github.com/midnightcodr/rpi_node_install
Now you can resume the "Install Gems" step in the Gitlab installation guide

sudo -u git -H bundle install --deployment --without development test postgres aws

and the rest

Notes

384MB is not enough to run "Compile assets" step on the gitlab installation guide so I had to add some more swap memory by following http://www.cyberciti.biz/faq/linux-add-a-swap-file-howto/

sudo dd if=/dev/zero of=/swapfile1 bs=1024 count=524288 sudo mkswap /swapfile1 sudo chmod 0600 /swapfile1 sudo swapon /swapfile1
Make sure the server_name setting in /etc/nginx/sites-available/gitlab matches gitlab_url in /home/git/gitlab-shell/config.yml, also add an entry to your RPI's /etc/hosts

127.0.0.1 gitlab.server.hostname
With the current version of Gitlab, performance is not that bad at all - it takes about 2 seconds to switch pages.

A few zsh tricks

Sat, 09 Nov 2013 22:53:00 GMT

Here are a few zsh tricks that I learned (and enjoy using) recently:

The basics - use !! to retrieve last command

tail /var/log/message
!!

The not so basic - get last parameter of last command with !$

mkdir new_folder
cd !$	# after this command you will be in folder new_folder

Get all parameters from the last command with !*

diff src/file1 dest/
cp !*

The pretty awsome - substitution with !!:s/from/to or !!:gs/from/to

	vi src/en/file.txt
	!!:s/en/fr

diff src/en/file.txt dest/en/file.txt
!!:gs/en/fr		# g for global substitution

Even better with ^

	vi src/en/file.txt
	^en^fr

	diff src/en/file.txt dest/en/file.txt
	^en^fr^:G

Switching between directories with similar structure (added Dec 7, 2013)

let's say you are in ~/dev/myproject1/src/lib/, and you want to cd to ~/dev/myproject2/src/lib/, all you need to do is

cd myproject1 myproject2

or even

cd 1 2

Batch renaming with zmv

	autoload zmv
	zmv '(*).txt' '$1.dat'	# change *.txt to *.dat

see more zmv examples at http://zshwiki.org/home/builtin/functions/zmv

Use && to simplify js codes

Sat, 20 Jul 2013 06:46:00 GMT

The technique will be illustrated by the following simple js codes:

{% include_code js-reg-and.js %}

Nagios audio alert with Mac OS X

Fri, 10 May 2013 00:50:00 GMT

Background

You have a nagios server running Linux and you have a nice OS at work called Mac OS X which comes with a handy text-to-speech feature. Wouldn't it be nice if you can turn your workstation into a nagios audio alert system? In this post I'll show you extactly how you can achieve that.

Requirements

node.js installed on the client, along with a module called execSync (installed by command npm install -g execSync)
nc (netcat) is installed on the nagios server (most distro comes with it so this shouldn't be a problem)

Code on the client

{% include_code nagios-audio-alert.js %} Run the code from terminal:

node /usr/local/bin/nagios-audio-alert.js

Do some test runs on the client

echo "service::PROBLEM::service down test"|nc -u -w 1 127.0.0.1 20123
echo "host::PROBLEM::host test123 is down"|nc -u -w 1 127.0.0.1 20123

If everything goes well, ^C to exit the nagios-audio-alert.js program and move on to nagios server.

Modification to nagios server's config files

command.cfg

	...
	define command{
			command_name    notify-host-by-tts
			command_line /usr/bin/printf "host::$NOTIFICATIONTYPE$::$HOSTNAME$ is $HOSTSTATE$"|nc -u -v -w 1 <replace_with_ip_of_mac_client> 20123
			}
	...

template.cfg

	...
	define contact{
			name                            generic-contact         ; The name of this contact template
			service_notification_period     24x7                    ; service notifications can be sent anytime
			host_notification_period        24x7                    ; host notifications can be sent anytime
			service_notification_options    u,c,r,f,s               ; send notifications for all service states, flapping events, and scheduled downtime events
			host_notification_options       d,u,r,f,s               ; send notifications for all host states, flapping events, and scheduled downtime events
			#service_notification_commands   notify-service-by-email        ; send service notifications via email
			service_notification_commands   notify-service-by-tts,notify-service-by-email	; notify thru audio & email
			#host_notification_commands      notify-host-by-email   ; send host notifications via email
			host_notification_commands      notify-host-by-tts,notify-host-by-email
			register                        0                       ; DONT REGISTER THIS DEFINITION - ITS NOT A REAL CONTACT, JUST A TEMPLATE!
			}
	...

Restart nagios server after making the above changes, always test nagios configuration before restarting: /etc/init.d/nagios checkconfig

Run nagios-audio-server.js as a daemon

Rason for this is that we want the Mac to be able to play alerts even when the user is not logged in. Create /System/Library/LaunchDaemons/nagios-audio-alert.plist with the following content, change CHANGE_TO_YOUR_USERNAME_ON_MAC to your mac username {% include_code nagios-audio-alert.plist.xml %} then

sudo launchctl load -w /System/Library/LaunchDaemons/nagios-audio-alert.plist

Some notes

20123 is the UDP port I grabbed from the air, change to whatever port you like (make sure they are in sync in both the js code and nagios config file command.cfg
Reason why execSync is needed is because if you have two (or more) messages come in at the same time (or almost simultaneously), you want the messages to be played one after another
If you have filewall on either the client or the nagios server, make sure they allow the traffic for the protocol/port used
A mac client is not really a hard requirement to build a nagios audio alert system like this, a Linux box that is capable of doing text-to-audio can handle this kind of task with ease, some code in nagios-audio-alert.js need to be changed though

GA-Z68X-UD3H-B3 F12 Mountain Lion 10.8.3 Installation Notes

Tue, 09 Apr 2013 00:02:00 GMT

Last Friday I had a hard time installing Moutain Lion 10.8.2 onto a GA-Z68X-UD3H-B3 (BIOS version F10), which was running fine with Lion (10.7.3), with either upgrade or clean install method. With some help from http://www.kakewalk.se/forums/discussion/4008/success-ga-z68x-ud3h-10-8-x-mountain-lion/p1 and http://www.tonymacx86.com/ I managed to install ML 10.8.3 onto the system. Here are the steps:

Make a 10.8.3 Unibeast bootable usb drive (doesn't need to check either options when making the drive with Unibeast)
Upgrade BIOS to F12 (tried UEFI version but got BIOS ID check error hence I settled with F12)
In BIOS setting, make sure HPET is set to 64bit. The system has an Agility 3 120GB SSD so the SATA3 port mode should be set to ACHI
After installation, boot from Unibeast drive again but choose the OS on the SSD
Download DDST (to the desktop), Multibeast from tonymacx86.com, choose the followings (More on the audio later) {% img center /images/z68x-ud3h-b3-f12-ML-10.8.3.png %}
After Multibeast and Lnx2Mac's Realtek driver installations, shutdown, remove Unibeast drive and boot from SSD, network should be working but audio is not working (can see the sound icon, can adjust volume but there's just no sound coming out)
Download Audio_Network.zip from http://ge.tt/6nuMCAL/v/0?c, unzip but only place AppleHDA.kext onto the desktop
Download KextBeast from tonymacx86.com and run it, it should pick up the AppleHDA.kext on the desktop
Run Multibeast again, select nothing but Drivers & Bootloaders->Drivers->Audio->Realtek ALC8xx->Without DSDT->ALC889 (basically the same option for audio in step 5)
Reboot, audio problem should now be fixed (try different line out port if is sound still missing).
(Optional) If you need to use iMessage, you need to install the newest Chimera (2.0.1 currently) from http://www.tonymacx86.com/downloads.php?do=file&id=164

other notes

Trim doesn't seem to be turned on when I checked the system info, need to spend some time to look into that.

Cool Audio Reminder script using MAC OS's text-to-speech

Sat, 26 Jan 2013 05:00:00 GMT

Why

We know that sitting for too long is harmful for our health. My solution to this problem is to set up a friendly reminder using Mac OS X’s speech to text feature, and a bit of programming. So here we go:

Steps

1). Create a reminder file under your home directory, for example,

$ vi ~/break_reminder

Stand up and walk around Your_Name.
Don't be lazy Your_Name, I know you can do it.
Stand up now, Your_Name, and do some exercises.
Sitting for too long is not healthy for you Your_Name.
Take a walk, get some water Your_Name.
You need a short break Your_Name.
Don't stick your butt to the chair for too long Your_Name.

Obviously change Your_Name to something that you want the voice to call you. This is one that I created, feel free to change the content.

2). Create the reminder shell script

$ vi ~/reminder.sh

#!/usr/bin/env bash
(( total_lines = $(wc -l < ~/break_reminder) )) && (( line = $RANDOM % $total_lines + 1 )) && sed -n ${line}p ~/break_reminder|say

3). Run some dry tests by running

bash ~/reminder.sh

4). Add the reminder to crontab

crontab -e

0 9-16 * * 1-5 bash ~/reminder.sh

This will run the reminder script from 9am to 4pm hourly on the clock Monday through Friday. If you are new to crontab, refer to http://en.wikipedia.org/wiki/Cron for more details.

Notes

I tried putting the meat inside ~/reminder.sh directly into crontab, didn’t work, that’s why this extra script is needed.
Feel free to change the env from bash to zsh in ~/reminder.sh, I have tested it to work with both bash and zsh.
Personally I prefer Serena’s voice, you need to download it via these steps: Search for “Dictation & Speech” in the spotlight -> Tick “Text to Speech” -> Tick “Customize”, then “Serena” from System Voice dropdown, you will be prompted to download the voice file if it has not been downloaded before.
The reminder (break_reminder) and script (reminder.sh) can be found in my github repository: https://github.com/midnightcodr/break_reminder.

Display Stand DIY

Mon, 31 Dec 2012 01:50:00 GMT

Ideas

Inspired by http://lifehacker.com/5872323/diy-ikea-monitor-stand-for-12, I decided to make a monitor stand that can support 2 LCD monitors.

I settled down on the following parts:

The boards: EKBY TRYGGVE softwood shelf
The legs: CAPITA 6" standless stell leg

Since one board needs 6 legs (I wouldn't try with 4 legs only) I ended up bying 2 boards and 3 packs of legs (12 legs total) so I can make an extra stand for my work place.

Final material cost (per stand): $25.

The making and the result

{% img https://dl.dropbox.com/u/16020214/monitor_stand/monitor_stand_in_the_making.jpg %} {% img https://dl.dropbox.com/u/16020214/monitor_stand/monitor_stand_final.jpg %}

Install node.js v0.8.14 onto Centos 5 howto

Fri, 23 Nov 2012 03:04:00 GMT

Installing through Linux binaries with the current newest version (v0.8.14) doesn't seem to work. Attempting to run "node -v" would give the folowing errors

node: /lib/libc.so.6: version `GLIBC_2.9' not found (required by node)
node: /lib/libc.so.6: version `GLIBC_2.6' not found (required by node)
node: /lib/libc.so.6: version `GLIBC_2.7' not found (required by node)

Therefore the only way to get it installed onto Centos 5.X (which comes with Python 2.4.3 which will fail while compiling node) is through compiling from source code. I am putting together this guide to ensure the installation is painlessly easy as I've been through the process a few times. The systems I've tested with happened to be both 32bit but I assume the procedures should work with the 64bit system as well, just make sure you download the source files matching the CPU architecture.

Steps

step 1. build bzip2 from source {% codeblock lang:bash %} wget http://bzip.org/1.0.6/bzip2-1.0.6.tar.gz tar xpzf bzip2-1.0.6.tar.gz cd bzip2-1.0.6 make make install {% endcodeblock %}

step 2. build Python 2.7 {% codeblock lang:bash %} wget -O - http://www.python.org/ftp/python/2.7.3/Python-2.7.3.tgz|tar xz cd Python-2.7.3 ./configure make make install {% endcodeblock %}

step 3. log out and log back in to the system

step 4. build node 0.8.14 {% codeblock lang:bash %} wget -O - http://nodejs.org/dist/v0.8.14/node-v0.8.14-linux-x86.tar.gz|tar xz cd node-v0.8.14 ./configure make make install {% endcodeblock %}

Notes

Step 1 is very critical for the installation of the node. Without it, you will get Python error "cannot find module bz2" while running make in step 4;
The solution provided by this guide will install Python 2.7 onto your system. If you found something broken with your Python programs, make sure you check the $PATH setting, the OS stock version should have python under /usr/bin/python while the compiled version should be /usr/local/bin/python. Adjust the orders of /usr/local/bin and /usr/bin in $PATh setting if needed.

References

http://stackoverflow.com/questions/812781/pythons-bz2-module-not-compiled-by-default http://nodejs.org/download/

Three ways to speed up Raspberry Pi

Mon, 12 Nov 2012 00:52:00 GMT

Raspberry Pi is a fun device to play with but sometimes we wish it can be a tad speedier. In this post I will show you how I did to speed up my rpi. I am running Debian on my rpi so some of the methods might not work for you if you are using different distro.

1. Overclock

With the newest Debian wheezy, simply run raspi-config, then overclock, OK, choose the one that works for you, choose OK. I settled with 900Mhz because once I use 950MHz, come tasks (such as compilings) could not run properly. Also check out http://www.raspberrypi.org/archives/tag/overclocking for more information on overclocking rpi.

2. Adjust memory split

By default I got only 184MB of usuable RAM for the system because the rest of the 256MB goes to GPU. While you are still at raspi-config's top menu, choose memory_split. If you run your rpi primarily as a headless server (not running X), just pick the one with the highest system RAM option (240), OK.

Note: method 1 and 2 require restart to take effect.

3. Find out auto-start services and de-select those that don't need to auto-start

For example, if you only need mysql server once a while, it makes no sense to make it run when rpi boots. A simple way to find out what processes are started during boot is through a program sysv-rc-conf (not installed by default). Install and fire it up through:

sudo apt-get install sysv-rc-conf
sudo sysv-rc-conf

Generate SSL certificate with 1 easy step

Thu, 11 Oct 2012 02:22:00 GMT

Originally I created a post on how to generate ssl certificate with one simple step: http://ricochen.wordpress.com/2010/01/01/generate-ssl-certificate-in-1-quick-step/

I am here to document the procedure in this first post of my Octopress based blog.

Command to generate SSL certificate (and its associated private key) is pretty straight-forward:

$ openssl req -new -x509 -days 3650 -keyout key.pem -out cert.pem -newkey rsa:2048 -subj "/CN=hostname.example.org"

If a password is created in the above step, you'll need one extra step to remove it unless you want to type the password everytime you restart the web server:

$ openssl rsa -in key.pem -out key.pem

Rico's Blog

The Hidden Danger of select on a Shared ioredis Connnection

The Setup

The Investigation

The Root Cause

The Fix

Takeaway

How I upgraded my home lab from Nginx to Caddy

References

How to wakeup/shutdown a PBS server from proxmox

Intro

Setup

Reference

How to handle file uploads using actix-web

Add error code(s) to Hapijs output

Step 1, assign error codes to each validation error

Step 2, customize failAction when creating Hapi server

Step 3, customize response

I just built my first standing desk

So it's for real

Here's the recipe

Note

Install Gitlab(6.4) on Raspberry PI

Steps

[update: added git-svn to the list on 1/17/2014]

be patient, the following command takes a while

and the rest

Notes

A few zsh tricks

The basics - use !! to retrieve last command

The not so basic - get last parameter of last command with !$

Get all parameters from the last command with !*

The pretty awsome - substitution with !!:s/from/to or !!:gs/from/to

Even better with ^

Switching between directories with similar structure (added Dec 7, 2013)

Batch renaming with zmv

Use && to simplify js codes

The technique will be illustrated by the following simple js codes:

Nagios audio alert with Mac OS X

Background

Requirements

Code on the client

Do some test runs on the client

Modification to nagios server's config files

Run nagios-audio-server.js as a daemon

Some notes

GA-Z68X-UD3H-B3 F12 Mountain Lion 10.8.3 Installation Notes

other notes

Cool Audio Reminder script using MAC OS's text-to-speech

Why

Steps

Notes

Display Stand DIY

Ideas

The making and the result

Install node.js v0.8.14 onto Centos 5 howto

Steps

Notes

References

Three ways to speed up Raspberry Pi

1. Overclock

2. Adjust memory split

3. Find out auto-start services and de-select those that don't need to auto-start

Generate SSL certificate with 1 easy step