all posts tagged LinuxCon Europe


by on October 28, 2015

Highlights from LinuxCon EU: Bitrot detection in GlusterFS

Some quick highlights from our Gluster talks at LinuxCon EU, as well as slides available as a PDF below.

Bitrot detection in GlusterFS  – Gaurav Garg, Venky Shankar

BitRot detection is a technique used to identify certain “insidious”
type of disk errors where data is silently corrupted with no
indication from the disk to the storage software layer that an error
has occurred. This class of data corruption are very different than
latent sector error typically caused by damaged disk drive. When data
gets “rotted”, an application can still access data off the disk drive
unknowing that the data is corrupted.

With GlusterFS 3.7, its possible to detect corruption caused due to
bitrot and take steps to rectify them. When bitrot detection is
enabled on a Gluster volume, files are signed after they have been
written. A periodic filesystem scrubber verifies the integrity of
signed files are flags (or marks) files which have mismatching
signature. Corrupted files are typically denied access to clients
unless it’s a replicated volume where it’s still possible to access
the “good” copy and repair the corrupted file.

LinuxCon EU – Bitrot Detection slides

by on October 27, 2015

Highlights from LinuxCon EU: Gluster Automatic File Replication Talk

Some quick highlights from our Gluster talks at LinuxCon EU, as well as slides available as a PDF below.

Gluster Automatic File Replication Talk from Ravishankar Narayankutty:

The talk started off with a quick-start introduction to the gluster lingo and its keywords, relating them to a block-diagram representation of the components.
It then went on to outline what a translator is and how AFR works on gluster, how files are healed, how a file can end up in split-brain etc. It then delved into
the recent improvements in AFR, namely:

1. Improvements in the ‘glfsheal’ program which gives out statistics for files that need heal.
2. The server and client side split-brain resolution commands that make it a breeze to resolve files that are in split-brain.
3. The ‘arbiter’ configuration for replica 3 volumes that prevent split-brain altogether.

Finally, the talk touched upon the upcoming improvements in self-heal performance, namely granular entry self-heals, granular data self-heals, mufti-threading the self-heal daemon, throttling and compounding of FOPS for better responsiveness and performance.

Slides for Gluster’s Automatic File Replication – LinuxCon EU