ClickHouse Migration Error (DB::Exception: Table system.crash_log doesn't exist)

Bug Description

This issue happens on fresh PostHog installation and when system.crash_log is absent on healthy ClickHouse instances
A fresh self-hosted PostHog installation can fail during ClickHouse migrations when using Docker.

There are two independent but related problems:

1. ClickHouse Docker Version Compatibility (AVX2 Requirement):
   Recent ClickHouse Docker images require AVX2-capable CPUs, preventing deployment on older hardware.
   During testing, the first issue encountered was unrelated to PostHog itself, but to the default ClickHouse Docker image used by the self-hosted deployment.
   At the time of writing, the master branch of PostHog uses the following default ClickHouse image:
   clickhouse/clickhouse-server:26.6.1.1193
   This image requires a CPU with AVX2 instruction set support. AVX2 CPU flag, the ClickHouse server fails immediately during startup, preventing the PostHog stack from completing initialization.
   After testing several releases, the newest version that successfully started on my hardware was:
   clickhouse/clickhouse-server:26.5.5.8
   Changing only the Docker image inside docker-compose.base.yml version allowed the ClickHouse container to start normally

2. Migration 0159 fails because system.crash_log is not created on a fresh, healthy ClickHouse instance

Migration posthog/clickhouse/migrations/0159_crash_log_metrics.py assumes system.crash_log already exists, but ClickHouse only creates this system table after a fatal server crash.
According to the ClickHouse documentation, system.crash_log is not created during installation or server startup.
Therefore, querying the table immediately after installation results in an exception similar to:

DB::Exception:
Table system.crash_log doesn't exist

Proposed Fix:
The attached migration applies the smallest possible change by disabling crash log metrics during the creation of the custom metrics view.
Original code :

CUSTOM_METRICS_VIEW(
    include_counters=True,
    include_server_crash=True,
)

Updated File (posthog/clickhouse/migrations/0159_crash_log_metrics.py):

from posthog.clickhouse.client.migration_tools import NodeRole, run_sql_with_exceptions
from posthog.clickhouse.custom_metrics import CUSTOM_METRICS_VIEW

operations = [
    run_sql_with_exceptions(
        CUSTOM_METRICS_VIEW(
            include_counters=True,
            include_server_crash=False,
        ),
        node_roles=[NodeRole.DATA],
    ),
]

Debug info

Start with posthog/clickhouse/migrations/0159_crash_log_metrics.py and inspect how CUSTOM_METRICS_VIEW handles the absent system.crash_log table. Review docker-compose.base.yml for the ClickHouse image version and verify both fresh migration behavior and the reported CPU compatibility issue; done means a fresh self-hosted installation completes migrations without requiring system.crash_log.