Histogram counter drops to 0 and then returns to the last value during vector pod restart

[ShukhratE]

Hi,
I am using prometheus_remote_write to send metrics.
During vector pods restart I see strange behavior when the histogram_counter metric is reset to 0 and then shows the last known state.

Any idea what can cause such an outcome?

Config of remote_write:

    [sinks.victoria_local]
      type = "prometheus_remote_write"
      inputs = ["vector_metrics_source"]
      endpoint = "http://10.187.0.47:8428/api/v1/write"
      buffer.type = "memory"
      buffer.max_events = 60000
      buckets = [ 0.005, 0.1, 0.5, 1, 3, 20 ]
      healthcheck.enabled = false
      default_namespace = "service" # required

[jszwedko]

This was discussed in Discord: https://discord.com/channels/742820443487993987/1060253332586893433/1060253332586893433