Archive for the ‘Kernel’ Category

SysDig for Docker & Kubernetes

28 Jan

The first thing the name “Sysdig” reminded me of was that of ht://Dig, for those of us who ever needed to implement some full text search on their website two decades ago, ht://Dig was the way to go. Now I am not sure if Sysdig guys had any inspiration from ht://Dig for the name, but I am sure it’s got some very interesting features that may well cause it to become something of a first choice for monitoring your containers.

So about a month ago, we had a demo by the SysDig guys over a conference call and I must say that I was impressed by the level of detail captured by this new monitoring solution. To sum it up in one line, it captures the type of system calls done by Docker containers (and hence your apps running in them) and Kubernetes clusters if you think on a broader scale: and lets you know precisely when things deviate from the learned or expected norms. I plan to run an evaluation soon and will share further experiences.

Read the rest of this entry »


Linux Kernel MTRR Trim problems

13 Mar

I recently came across an interesting problem with a newly purchased set of servers. These were HP DL165 G7, very nice 1U machines with dual sockets and capable of good RAM expansion. Once the OS (64 Bit CentOS) was booted it couldn’t see the full 16GB of installed memory. The memory being reported was only 3011MB, 3GB approx.

On closer examination it was established that the dmesg log had a indicated a potential problem, the message suggested that some of the memory, although detected by the kernel was “trimmed” and the box was left with 3GB to run.

The problem was fixed by enabling the kernel argument “disable_mtrr_trim”.

No Comments

Posted in Kernel, Linux