Automatically split the payload for metrics based on the size of the payload before sending to EEC

[j01am]

We get a lot of errors like : code: 413 message: Payload must not be larger than 1048576 bytes Would it be beneficial to implement split sending based on size, in addition to splitting by the number of lines?

[dlopes7]

Maybe, this is also more resource intensive

Can you please share some examples of these long lines you need to ingest?

You can also split the metrics before reporting them

[j01am]

Hi !

We are hitting the size limit with Metrics that have around 25 dimensions. A potential simple solution could be to address the issue with a try catch : If a size exception is caught, you can recursively call the send function with half of the payload.

For example, we sent lines with more than 1100 char :

send_metric: cloud.alicloud.ecs.memory_utilization,alicloud.instance_name="XXXXXX",alicloud.instance_id="i-gw82jct5y8ey92mwso2w",alicloud.instance_resource_group_id="",alicloud.instance_memory="8192",alicloud.instance_charge_type="PrePaid",alicloud.instance_cpu="2",alicloud.instance_network_type="vpc",alicloud.instance_host_name="XXXXXX",alicloud.instance_vlan_id="",alicloud.instance_stopped_mode="Not-applicable",alicloud.nstance_deletion_protection="False",alicloud.instance_internet_charge_type="PayByTraffic",alicloud.instance_deployment_set_id="",alicloud.instance_serial_number="xxxxx-xxxxxx-xxxxx-xxxxx",alicloud.instance_os_type="windows",alicloud.instance_gpu_spec="",alicloud.instance_device_available="True",alicloud.instance_type="ecs.sn2ne.large",alicloud.instance_os_name_en="Windows Server 2019 Data Center Edition 64bit English Edition",alicloud.instance_zone_id="eu-central-1a",alicloud.instance_cluster_id="",alicloud.instance_gpu_amount="0",alicloud.instance_region_id="eu-central-1",alicloud.instance_status="Running",alicloud.instance_networkinterfaceip="xx.xx.xx.xx" gauge,59.26 1716801540000