So I have a 3 node nebula deployment, and each sto...
# nebulaj
Jeremy Simpson
12/12/2024, 12:28 AMSo I have a 3 node nebula deployment, and each storaged pod takes ~20Gb of memory. This is even if I have zero requests. What is taking up all the memory? I'm trying to figure out what I can tweak to reduce it. Is it rocksdb? Is it mostly tied to the raft groups? Thanks!
d
Dinah
12/12/2024, 5:50 AMHi, could you provide the version of NebulaGraph currently in use? Additionally, I would like to inquire about the configuration settings for the block cache size and any specific operations or procedures that have been carried out regarding this setting.
j
Jeremy Simpson
12/13/2024, 10:24 PMI am using pretty much default configs. I use the k8 operator. Version 3.6.0. Also, it's been running for awhile now -- I have a few graph spaces, so the db isn't empty per se
Jeremy Simpson
12/13/2024, 10:25 PMEven if I restart the pods, it takes up ~20GB on startup with zero requests -- if I give it less memory than that it crashes with OOMKilled a few moments after it starts up. So it leads me to think some cache or something is being loaded into memory on startup
Jeremy Simpson
12/13/2024, 10:26 PMI'm wondering what that could be and if I could reduce it -- or am I SoL because it's tied to something I can't really control
Jeremy Simpson
12/13/2024, 10:26 PMThank you!
6 Views