Pattern_ingester persistence

Grafana Loki
1

Hi,

I’m using the Loki helm chart and have enabled the pattern_ingester. It looks to work fine (patterns get displayed in Grafana) but I’m just wondering where the pattern_ingester would persist its state. There is a persistence config within the helm chart but the created volume actually stays empty. Every pod restart resets the patterns / discards the old ones.

Helm config:

...
                    'loki': {
                        ...
                        'pattern_ingester': {
                            'enabled': True,
                        },
                    ...
                    'patternIngester': {
                       ...
                        'persistence': {
                            'enabled': True,
                            'claims': [
                                {
                                    'name': 'data',
                                    'size': '1Gi',
                                },
                            ],
                        },

Note that I’m running on Istio with very restrictive settings and I had to make some changes that the distributor was able to call the pattern_ingester (for some reason the helm chart only deploys a StatefulSet without any k8s services…), so maybe there is another culprit. But could not spot any error logs in Loki or failed connects in the istio sidecar proxies as of now.

Any ideas/hints?