> For the complete documentation index, see [llms.txt](https://docs.p0.dev/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.p0.dev/agentic-access/agentic-gateway/log-provider/victorialogs.md). # VictoriaLogs Read gateway activity from the VictoriaLogs instance the batteries-included chart deploys in your cluster. The **VictoriaLogs** log provider reads your gateway's MCP and A2A activity from [VictoriaLogs](https://docs.victoriametrics.com/victorialogs/), a log store that the batteries-included chart deploys in your cluster next to the gateway (see [Install the batteries-included Helm chart](/agentic-access/agentic-gateway/gateway/deploy/install-the-batteries-included-chart.md)). The chart and the console installer ([Install from the P0 console](/agentic-access/agentic-gateway/gateway/deploy/install-from-the-p0-console.md)) both turn it on by default, so a gateway installed with either one is already writing its activity there. The VictoriaLogs provider is in beta. ## How it works The gateway sends its logs to an OpenTelemetry collector in the same release, which writes them to VictoriaLogs. VictoriaLogs runs as an internal cluster service with no authentication of its own, and nothing outside the cluster can reach it. P0 reads activity through the gateway's management API, with the same access it already uses to manage the gateway: the service account the gateway trusts through `accessProxy.manageAllowedEmails`. You don't create any credentials, DNS records, or open ports for reading logs. ## Prerequisites * A registered [Gateway](/agentic-access/agentic-gateway/gateway.md) component, deployed as in [Install the batteries-included Helm chart](/agentic-access/agentic-gateway/gateway/deploy/install-the-batteries-included-chart.md) or [Install from the P0 console](/agentic-access/agentic-gateway/gateway/deploy/install-from-the-p0-console.md). * The chart's defaults for VictoriaLogs left in place: | Value | Default | | ---------------------------------------- | -------------------------- | | `victorialogs.enabled` | `true` | | `collector.enabled` | `true` | | `ai-gateway.accessProxy.victorialogsUrl` | `http://victorialogs:9428` | | `ai-gateway.observability.otlpEndpoint` | the bundled collector | If any of these were changed to turn logging off, set them back and upgrade the release. ## Add the log provider 1. Navigate to **Integrations** on [p0.app](https://p0.app) and select **Agentic gateway**, then choose the **Log provider** component. 2. Click **Add log provider**, and enter a **Log provider identifier**. 3. Set **Provider** to **VictoriaLogs (via gateway)**. There is nothing else to fill in. 4. Link each gateway to the new provider, as in [Log provider](/agentic-access/agentic-gateway/log-provider.md#link-a-gateway-to-a-log-provider). One VictoriaLogs provider serves every gateway you link to it. ## Retention and storage VictoriaLogs keeps activity for 30 days by default. To change that, set `victorialogs.retentionPeriod` to a value such as `7d`, `90d`, or `1y`. Its volume is 20Gi by default, set by `victorialogs.storageSize`. On the default persistent volume, VictoriaLogs removes logs only by age, not by size, so size the volume to hold a full `retentionPeriod` of logs. See [Deploying the gateway](/agentic-access/agentic-gateway/gateway/deploy.md#cluster-size) for the chart's other volumes. {% hint style="warning" %} With `victorialogs.persistence.enabled: false`, logs are written to scratch space and are lost whenever the VictoriaLogs pod is rescheduled. Use that setting only to evaluate the gateway on a cluster with no `StorageClass`. In this mode, `storageSize` caps the scratch space, and VictoriaLogs also drops its oldest logs once they fill 80% of it. VictoriaLogs always keeps the two newest days, so size the scratch space to fit two days of logs in 80% of it. {% endhint %} The bundled VictoriaLogs is a single instance with no replication or backups. For high log volume, run a clustered VictoriaLogs yourself, point the collector at it, and set `ai-gateway.accessProxy.victorialogsUrl` to its address. The gateway must be able to reach that address from inside the cluster. ## What the Event log shows By default, the [Event log](/agentic-access/using-the-gateway/event-log.md) shows the last 7 days of activity. Pick a time range to see older entries. P0 returns one page of the most recent matching entries across your gateways and doesn't page further back, so narrow the time range or add filters to find older calls. ## Troubleshooting ### The Event log shows an error instead of activity P0 couldn't query a gateway. P0 queries every linked gateway for each page, so one unreachable gateway turns the whole Event log into an error, not just that gateway's rows. Check each gateway linked to the provider. First check that the gateway is reachable from P0, as described in [Gateway](/agentic-access/agentic-gateway/gateway.md#the-install-fails-with-a-reachability-error). Then check that the gateway knows where VictoriaLogs is. If `victorialogsUrl` is empty, the gateway answers the query with `404`: ```bash kubectl -n get deploy agentic-gateway-server \ -o jsonpath='{.spec.template.spec.containers[0].env[?(@.name=="VICTORIALOGS_URL")].value}' ``` An empty result means `ai-gateway.accessProxy.victorialogsUrl` isn't set. Set it to `http://victorialogs:9428` and upgrade the release. If `victorialogsUrl` is set but VictoriaLogs isn't running, the gateway answers with `502` instead. This happens when `victorialogs.enabled` is `false` but `victorialogsUrl` still has its default. Turn VictoriaLogs back on, or point `victorialogsUrl` at the VictoriaLogs instance you run yourself. ### No activity appears, and there is no error The query worked but found nothing. Confirm the collector and VictoriaLogs are running: ```bash kubectl -n get pods | grep -E 'otel-collector|victorialogs' ``` Both should read `Running`. If they do, make a tool call through the gateway, then refresh the Event log. If the call still doesn't appear, check that `ai-gateway.observability.otlpEndpoint` points at the bundled collector and hasn't been cleared.