Open wangting0128 opened 1 month ago
/assign @aoiasd /unassign
May relate https://github.com/milvus-io/milvus/issues/25767 How about the pulsar conifg
Milvus-Helm don't change the setting of backlogQuoteDefaultLimitGB Seems some dead subdescribe cause pulsar backlog larger than limit, and cause create producer panic.
the cpu keeps increasing and memory usage is also increaasing.
My guess;
problem is:
Is there an existing issue for this?
Environment
Current Behavior
argo task: fouramf-8276l
server:
enabled global mmap
proxy panic holiday-mmap-milvus-proxy-65b4cc7c8d-gbw2r_panic.log
client log:
Expected Behavior
No response
Steps To Reproduce
Milvus Log
No response
Anything else?
server config:
test result: