metrics after 12 hours at rest with no collections/documents { “system_cpu1_active_percentage”: “0.00", “system_cpu2_active_percentage”: “0.00", “system_cpu3_active_percentage”: “0.00", “system_cpu4_active_percentage”: “0.00", “system_cpu5_active_percentage”: “0.00", “system_cpu6_active_percentage”: “0.00", “system_cpu7_active_percentage”: “0.00", “system_cpu8_active_percentage”: “10.00", “system_cpu_active_percentage”: “2.47", “system_disk_total_bytes”: “85878374400", “system_disk_used_bytes”: “656809984", “system_memory_total_bytes”: “16439255040", “system_memory_total_swap_bytes”: “2147479552", “system_memory_used_bytes”: “1292021760", “system_memory_used_swap_bytes”: “0", “system_network_received_bytes”: “187676672", “system_network_sent_bytes”: “28823607", “typesense_memory_active_bytes”: “2478309376", “typesense_memory_allocated_bytes”: “2355328416", “typesense_memory_fragmentation_ratio”: “0.05", “typesense_memory_mapped_bytes”: “2558328832", “typesense_memory_metadata_bytes”: “162131080", “typesense_memory_resident_bytes”: “2478309376", “typesense_memory_retained_bytes”: “283312128" }%

it’s a node with 16gb of memory, 8 vcpu

I did have a collection in it previously, but after I deleted it and the memory refused to fully release I downgraded and then re-upgraded the node capacity. The metrics above are after the node was re-upgraded and at rest with no collections for >12 hours. Also, interestingly, the metrics above don’t seem to line up with the memory usage shown in the Cloud graph screenshot here, though maybe I’m mistaken 🤔 I’m using typesense version 27.1

when I create a brand new instance with the same configuration, it uses significantly less memory { “system_cpu1_active_percentage”: “0.00", “system_cpu2_active_percentage”: “0.00", “system_cpu3_active_percentage”: “0.00", “system_cpu4_active_percentage”: “0.00", “system_cpu5_active_percentage”: “0.00", “system_cpu6_active_percentage”: “0.00", “system_cpu7_active_percentage”: “0.00", “system_cpu8_active_percentage”: “0.00", “system_cpu_active_percentage”: “0.00", “system_disk_total_bytes”: “85857402880", “system_disk_used_bytes”: “650887168", “system_memory_total_bytes”: “16439259136", “system_memory_total_swap_bytes”: “2147479552", “system_memory_used_bytes”: “984768512", “system_memory_used_swap_bytes”: “0", “system_network_received_bytes”: “113035983", “system_network_sent_bytes”: “1761903", “typesense_memory_active_bytes”: “350879744", “typesense_memory_allocated_bytes”: “243142208", “typesense_memory_fragmentation_ratio”: “0.31", “typesense_memory_mapped_bytes”: “424411136", “typesense_memory_metadata_bytes”: “34095648", “typesense_memory_resident_bytes”: “350879744", “typesense_memory_retained_bytes”: “43778048" }%

at the time of the metrics/screenshot

with no change over time (that I can see)

ah actually, it’s slightly different - it’s using the default

num_dim

and

openai/text-embedding-3-small

. Otherwise it’s the same