Queue Groups on leaf clusters not balancing correctly when messages are routed in from hub cluster [v2.10.22] - Variant 2 #6040
Assignees
Labels
defect
Suspected defect such as a bug or regression
Comments
|
@roeschter I believe I fixed the issue in #6043. If you want to give it a shot. |
derekcollison
added a commit
that referenced
this issue
Oct 25, 2024
…des (#6043) When receiving a message from a route, if the queue interest is only routes and leaf connections, the server will always favor the leaf kind. However, if there were more than one leaf connections for the same queue group, the server will always pick the same (the last one). It would change if new leaf were to connect or disconnect/reconnect, but with a stable set, it would always be the same. This PR makes this selection somewhat random. Resolves #6040 Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
|
Question from end user. Whats the approximate release of this in 2.10? (2.11.23?) |
neilalexander
pushed a commit
that referenced
this issue
Oct 29, 2024
When receiving a message from a route, if the queue interest is only routes and leaf connections, the server will always favor the leaf kind. However, if there were more than one leaf connections for the same queue group, the server will always pick the same (the last one). It would change if new leaf were to connect or disconnect/reconnect, but with a stable set, it would always be the same. This PR makes this selection somewhat random. Resolves #6040 Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
neilalexander
pushed a commit
that referenced
this issue
Oct 29, 2024
When receiving a message from a route, if the queue interest is only routes and leaf connections, the server will always favor the leaf kind. However, if there were more than one leaf connections for the same queue group, the server will always pick the same (the last one). It would change if new leaf were to connect or disconnect/reconnect, but with a stable set, it would always be the same. This PR makes this selection somewhat random. Resolves #6040 Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Observed behavior
Queue Groups on leaf clusters not balancing (only one subscriber receiving messages) in certain assymetrically connected leaf cluster
Expected behavior
Queue Groups always balance when all subs connection in the same leaf cluster
See also: #5972
Server and client version
Server 2.10.22
nat-cli from main
Host environment
Windows (Michael) - Linux (customer)
Steps to reproduce
Set up a hub cluster with nodes HUB1, HUB2, HUB3
Set up a leaf cluster with nodes LEAF1, LEAF2, LEAF3
Connect the leaf nodes as follows: (LEAF1 and LEAF2 connect to the SAME Hub - this is NOT a typo)
LEAF1 → HUB1
LEAF2 → HUB1
LEAF3 → HUB3
Start queue group listeners on LEAF1 and LEAF2
nats --context LEAF1 sub --queue=q1 foo
nats --context LEAF2 sub --queue=q1 foo
Publish to HUB3
nats --context HUB3 pub foo Hello
The text was updated successfully, but these errors were encountered: