Replies: 1 comment
-
Any idea on this? Been running a couple tests with no desired result based on latency etc (thought it'd read first from less latency node all times but seems random) |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello everyone!
I've been tinkering with gluster last month and I'm doing some tests etc.
I have the next environment:
Nodes: A,B,C,D
Clients: M1 and M2
Volume is replicated-distributed (replica 2), so A/B are mirrors and C/D are also mirrors from one to another.
Nodes A ,C and client M1 are in Europe, whereas nodes B, D and client M2 are in America (I know about latency issues and what not but let's ignore that)
From the tests I've been running, it seems that clients read from nodes indiscriminately, Is there a way to specify that they should read from the node with less MS? So client M1 will read from nodes A and C and client will read from nodes B and D. Of course if by any chance any node goes down, it should read from the other one.
Sorry for my english and thank you for your answers
Best regards
EDIT: I know there's a tag cluster.read-subvolume-index, but that would make all clients to target 1 node based on index number, not latency
Beta Was this translation helpful? Give feedback.
All reactions