[cs615asa] Meetup Summary: Troubleshooting Containers & Kubernetes at Buffer Update

[Zhitao Chen]

[cs615asa] Meetup Summary: Troubleshooting Containers
& Kubernetes at Buffer Update

Topic:

Troubleshooting Containers & Kubernetes at Buffer
Update

Date:

March 27th, 2018

Link:

https://www.meetup.com/New-York-Kubernetes-
Meetup/events/248205124/

The reason that took this meetup:

The meetup is mainly about how to use their open source platform -- Sysdig
to troubleshooting. And we just did our assignment about using tcpdump(1)
to obverse the traffic, and tcpdump(1) is very useful for troubleshooting
the network traffic but it has some disadvantage like cannot highlight the
network traffic and make people easy to see what is going on.So I want to
know more about that field and see if there are another tools to do that
job.

About the meetup:

The meetup is contained in two talks, the first talk is held by Michael
Ducy, Director of Community & Evangelism for Sysdig. The second talk is
held by Dan Farrelly, Director of Engineering and Technical Lead at Buffer.

In the first talk, Michael Ducy introduced Sysdig and gave some example to
how to use it. And there are a few points that I think it is interesting:

1. Sysdig enables real-time and offline troubleshooting. Also, Sysdig is
pretty easy to use since its usage is similar to tcpdump(1). And
surprisingly the output of Sysdig is similar to tcpdump(1) also.

2. In the platform, there is a tool named Sysdig
Chisels which similar to Wireshark, it provides the
GUI to let the users observe the files R/W, network traffic, OS states etc.
As a result, it easy to observe different kind of data/state, e.g.
non-device files written in /dev; process tries to access camera; overwrite
system binaries; etc.

3. After all, I feel disappointed about this tool. It just combines
something together, and I cannot see something which Sysdig better than
tcpdump(1) or Wireshark.

The second talk is talk about the log of Buffer company using Kubernetes.

1. Why Buffer uses Kubernetes is because of its reliability, and also it
can increase developer velocity.

2. If we want to start Kubernetes, maybe it is a good start to lean Vanilla
Kubernetes first and pay attention to resource limits.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.stevens.edu/pipermail/cs615asa/attachments/20180405/79187547/attachment.html>