Analyze http2 with Wireshark

Most of the time http2 requests are sent over TLS, which means that Wireshark needs to decrypt information before we can analyze it. Fortunately, Chrome allows to dump cryptographic keys in the file. To do that, before you start Chrome, export environmental variable SSLKEYLOGFILE:

export SSLKEYLOGFILE=/home/jsn/Downloads/sslkeys/sslkeys.log

Firefox also allows to do it, but as for now (48) it is disabled by default. Get more information Now we need to configure Wireshark to use exported keys. Go to Edit - Preferences - Protocols - SSL and point (Pre)-Master-Secret log filename to the file with keys (/home/jsn/Downloads/sslkeys/sslkeys.log) Now you can filter records by http2 and analyze them.

tar: Archives cheatsheet

Create archive:

tar -czvf compressed_file.tar.gz original_file.sql

Extract files:

tar -xzvf compressed_file.tar.gz

Number of connections to the specific port

To check how many connections are exists to the port 3306:

watch -n 1 "sudo netstat -anp | grep 3306 | wc -l"

fedora: Increase watches limit

Open configuration file:

sudo vim /etc/sysctl.d/90-override.conf

Add the following line to the end of the file to increase watches limit:

fs.inotify.max_user_watches = 524288

postgresql: How to fix ERROR: new encoding (UTF8) is incompatible

Sometimes I'm unable to create a database in UTF8 encoding because the template database is in the ASCII encoding. To fix it we need to recreate template database in UTF8. Start psql console:

psql -U postgres

Run the following commands:

UPDATE pg_database SET datistemplate = FALSE WHERE datname = 'template1';

DROP DATABASE template1;

CREATE DATABASE template1 WITH TEMPLATE = template0 ENCODING = 'UNICODE';

UPDATE pg_database SET datistemplate = TRUE WHERE datname = 'template1';

\c template1

VACUUM FREEZE;

Preview apiary.io docs

It is possible to preview apiary docs before pushing to the repository. To do that, install apiaryio CLI:

gem install apiaryio

The executable file of the installed application is located somewhere there:

/home/jsn/.gem/ruby/gems/apiaryio-0.3.3/bin/apiary

Generate html file to preview docs and open it in the browser:

/home/jsn/.gem/ruby/gems/apiaryio-0.3.3/bin/apiary preview --output="preview.html"

fedora: Install HipChat 4

UPDATED: Atlassian team has a rpm repository which they do not mention on the official download page. To install hipchat4 from it, create a repository file:

sudo gedit /etc/yum.repos.d/hipchat4.repo

With the following content:

[atlassian-hipchat4]
name=Atlassian Hipchat4
baseurl=https://atlassian.artifactoryonline.com/atlassian/hipchat-yum-client/
enabled=1
gpgcheck=0

Install hipchat:

sudo dnf install hipchat4 --refresh

-----
Atlassian team wasn't able to release rpm package for the latest hipchat for more than 2 months. Fortunately, it is possible to intsall hipchat from the debian package. Download the latest available debian package from here.

Install alien package which will convert deb package to the rpm package:

sudo dnf install alien

Convert hipchat debian package to rpm package:

sudo alien HipChat4-4.0.1635-Linux.deb -r

Install hipchat:

sudo dnf install hipchat4-4.0.1635-2.x86_64.rpm

fedora: X11 forwarding issue

Here I have a nice post how to open browser window on the remote machine. On fedora 23 that method stopped to work. It was possible to start simple application like xclock, but firefox or chrome.
The problem is in the xorg-x11-drv-intel package. The latest version which is available in the official fedora repo is xorg-x11-drv-intel-2.99.917-19.20151206.fc23 (as for 13 May 2016). Downgrading the package to the previous version fixes the issue.
The latest version of the package which is known to work is xorg-x11-drv-intel-2.99.917-19.20151206.fc23. Let's install it:

sudo dnf install xorg-x11-drv-intel-2.99.917-16.20150729.fc23.x86_64

It would be also nice to lock the package version, so the next time you start updating the system it won't be updated:

sudo dnf install 'dnf-command(versionlock)'
sudo dnf versionlock add xorg-x11-drv-intel

How to format USB stick as FAT32

First step is to find the name of the USB stick in the system:

sudo fdisk -l

In my case the name of the device is /dev/sdc If the USB stick is mounted, unmount it with command:

sudo umount /dev/sdc

Finally, run the command to format USB stick:

mkdosfs -F 32 -I /dev/sdc

apt-get: Install package with specific version and hold it on the next upgrade

If you need to install specific version of the package in your system, you can list all available package versions with the command:

apt-cache madison varnish

After you know the package version you can install it with:

sudo apt-get install varnish=4.0.2-1

To hold the package on the selected version during the next upgrade:

sudo apt-mark hold varnish

You can also "unhold it" with:

sudo apt-mark unhold varnish

fedora: Install accidentally removed sqlite package

If sqlite package (or, for example, sqlite-libs package) was accidentally removed from your system, you can't reinstall it using dnf or yum because they depend on it. To fix this problem, download missing rpm package from https://admin.fedoraproject.org/pkgdb/package/rpms/sqlite/. From the downloaded file location, run command:

rpm2cpio sqlite-libs-3.11.0-3.fc23.x86_64.rpm | cpio -idmv

It will create usr folder with extracted files. Run next command to copy them to the right location:

sudo cp -R usr/ /

Pay attention to the package architecture. You should be able to use dnf again. It would be good to reinstall missed package with dnf again:

sudo dnf --best --allowerasing install sqlite-libs

Set up logging with Flask and gunicorn

Let's say that we want to be able to log in the specified log file and syslog. Our application - is a simple app.py file. The first step, I suggest, is to set logger name:

import os

from flask import Flask


app = Flask(__name__)
app.logger_name = "flask.app"

# ...

if __name__ == "__main__":
    app.run()

You can log in your project like this:

app.logger.info("Hello")

A log configuration file for the gunicorn may look something like this:

[loggers]
keys=root, gunicorn.error, gunicorn.access, requests.packages.urllib3.connectionpool, __main__

[handlers]
keys=log_file, syslog

[formatters]
keys=generic

[logger_root]
level=INFO
handlers=log_file, syslog

[logger___main__]
level=DEBUG
handlers=log_file, syslog
propagate=0
qualname=__main__

[logger_gunicorn.error]
level=INFO
handlers=log_file, syslog
propagate=0
qualname=gunicorn.error

[logger_gunicorn.access]
level=INFO
handlers=log_file, syslog
propagate=0
qualname=gunicorn.access

[logger_requests.packages.urllib3.connectionpool]
level=WARN
handlers=log_file, syslog
propagate=0
qualname=requests.packages.urllib3.connectionpool

[handler_syslog]
class=logging.handlers.SysLogHandler
formatter=generic
args=()

[handler_log_file]
class=logging.FileHandler
formatter=generic
args=('/home/myuser/log/gunicorn.log',)

[formatter_generic]
format=%(asctime)s [%(process)d:%(name)s:%(lineno)s] [%(levelname)s] %(message)s
datefmt=%Y-%m-%d %H:%M:%S
class=logging.Formatter

The last step is starting gunicorn:

gunicorn app:app --log-config /home/myuser/gunicorn_logging.conf

Conditions in the Vagrantfile

Every developer could have different hardware configuration. Let's say one is ready to provision a virtual machine with 4 cpus and another - only with 2 cpus. It would be nice to handle this situation with environmental variables:

Vagrant.configure(2) do |config|
  config.vm.box = "box-cutter/debian81"

  config.vm.network "private_network", ip: "192.168.33.10"
  config.ssh.forward_agent = true

  config.vm.provider "virtualbox" do |vb|
    vb.memory = "4096"
    if ENV.has_key?('VAGRANT_CPUS')
      vb.cpus = ENV['VAGRANT_CPUS']
    else
      vb.cpus = "2"
    end
  end
end

With that config the number of the provisioned cpus can be controled via VAGRANT_CPUS environmental variable.

Gnome: How to record a screencast

Gnome 3 has in-built screencast functionality. To start record your screen simply press

Ctrl + Shift + Alt + R

The indicator (a red circle) will appear in the panel, informing you that the recording has started. Default video length is limited to 30s, what might be not enough in most cases. To adjust this parameter install dconf Editor and go to the:

org.gnome.settings-daemon.plugins.media-keys

And look for max-screencast-length to adjust.

komodo: Use custom version of the jshint

Komodo has an outdated version of the jshint. If you want to use, for example, moz:true option, you need to upgrade it. I tried it multiple times and never succeed. The problem was not in komodo (at least that was the developers say) but in jshint. All latest versions of the jshint does not work with komodo. The latest version of the jshint that is known to work with komodo is 2.6.0. To configure komodo to use jshint 2.6.0, open File -> Preferences -> Syntax Checking -> Language: Javascript and point Custom linter to the downloaded file:

The version 2.6.0 is somewhere from the early 2015 but still much better then the default one.