Redeploy Tyk Pump lost previous data

Hi,

I deploy the Tyk Pump in pod in k8s and use the Prometheus to scrape it. When I redeploy the Pump, all the previous data are lost.
For example, I send one request and the Prometheus records there is one request. Then I redeploy the Pump and send one request. The Prometheus shows there is still one request instead of two. Is there a way to keep the data?

Here’s my config:

{
  "analytics_storage_type": "redis",
  "analytics_storage_config": {
    "type": "redis",
    "host": "redis",
    "port": 6379,
    "username": "",
    "password": "",
    "database": 0,
    "optimisation_max_idle": 2000,
    "optimisation_max_active": 4000
  },
  "purge_delay": 2,
  "pumps": {
    "prometheus": {
      "type": "prometheus",
      "meta": {
        "listen_address": ":8084",
        "path": "/metrics",
        "track_all_paths": true,
        "custom_metrics":[
          {
            "name":"proxy_http_requests_per_path",
            "description":"Total of API requests per path",
            "metric_type":"counter",
            "labels":["response_code","api_name","method","path"]
          },
          {
            "name":"proxy_http_requests_per_user",
            "description":"Total of API requests per user",
            "metric_type":"counter",
            "labels":["response_code","api_name","method","path","alias"]
          },
          {
            "name":"proxy_http_latency",
            "description":"Latency of API requests",
            "metric_type":"histogram",
            "labels":["type","response_code","api_name","method","path"]
          }
        ]
      }
    }
  },
  "dont_purge_uptime_data": true,
  "statsd_connection_string": "graphite:8125"
}

Hi,

Pump doesn’t store data at all. It picks up analytics records and “pumps” them on to various log aggregating tools. Once the analytics have left pump it is not able to affect the retention of the data within those tools.

We do have many customers who use Prometheus and don’t experience the issue you’re seeing so we know it’s possible, but unfortunately we don’t have the Prometheus expertise to tell you how that’s done beyond assuring you that it is possible.

Sorry we couldn’t be of more help

Cheers,
Pete