Parameter Configuration

How to skip the AVX2 CPU check at install time, and after install how to change the login account, configure Doris connection credentials and FE/BE addresses, tune Ingest pipeline task settings, configure Trace resource ignore filtering, SQL normalization, and adjust telemetry retention. For directories and lifecycle, see Docker Operations and Kubernetes Operations. For capacity planning and additional knobs, see Performance Tuning.

The open-source Web uses a built-in seed account loaded at startup (in memory). There is no UI to change credentials. Override via environment variables and restart ai-apm-web.

YAML key	Environment variable	Default
`apm.security.seed-username`	`APM_SECURITY_SEED_USERNAME`	`admin`
`apm.security.seed-password`	`APM_SECURITY_SEED_PASSWORD`	`Databuff@123`

Sign in again with the new account afterward; old sessions/tokens will not work. The password is stored in env vars—protect host access accordingly. After start.sh / install completes, the summary prints the effective credentials from the running Web container (not a hard-coded default).

Docker

Preferred: docker-compose.override.yml (merged automatically; upgrades do not overwrite it):

bash

cd /opt/databuff-ai-apm   # or: echo $APM_INSTALL_DIR

cat > docker-compose.override.yml <<'EOF'
services:
  ai-apm-web:
    environment:
      APM_SECURITY_SEED_USERNAME: "youruser"
      APM_SECURITY_SEED_PASSWORD: "your-strong-password"
EOF

docker compose up -d ai-apm-web
docker exec ai-apm-web printenv | grep '^APM_SECURITY_SEED_'

You can edit docker-compose.yml under ai-apm-web.environment instead, but upgrades replace docker-compose.yml—re-apply changes or use override.

Kubernetes

Add keys to the ConfigMap and restart Web:

bash

kubectl -n databuff edit configmap ai-apm-config

Under data::

yaml

  APM_SECURITY_SEED_USERNAME: "youruser"
  APM_SECURITY_SEED_PASSWORD: "your-strong-password"

Then:

bash

kubectl -n databuff rollout restart deploy/ai-apm-web
kubectl -n databuff rollout status deploy/ai-apm-web
kubectl -n databuff exec deploy/ai-apm-web -- printenv | grep '^APM_SECURITY_SEED_'

Editing the Deployment env (kubectl -n databuff edit deploy/ai-apm-web) works the same way.

Upgrade note: Re-running install.sh / start.sh applies the stock configmap.yaml and may wipe hand-added keys. Re-apply them after upgrade, or keep a local patch checklist.

2. Configure Doris username and password

Credentials used by Web / Ingest for Doris JDBC and Stream Load (not the Web UI login above). Defaults match a stock Doris install: root with an empty password.

Component	YAML key	Environment variable	Default
Web	`apm.doris.username`	`DORIS_USER`	`root`
Web	`apm.doris.password`	`DORIS_PASSWORD`	empty
Ingest	`ingest.doris.username`	`DORIS_USER`	`root`
Ingest	`ingest.doris.password`	`DORIS_PASSWORD`	empty

Both components share the same env vars; restart ai-apm-web and ai-apm-ingest after changing them.

Order: Change or create the Doris account first (e.g. SET PASSWORD FOR 'root' = PASSWORD('your-strong-password');), then update the app env vars. Changing only the app side while Doris still uses an empty password will break connectivity.

Docker

Prefer docker-compose.override.yml (set on both Web and Ingest):

bash

cd /opt/databuff-ai-apm   # or: echo $APM_INSTALL_DIR

cat > docker-compose.override.yml <<'EOF'
services:
  ai-apm-web:
    environment:
      DORIS_USER: "root"
      DORIS_PASSWORD: "your-strong-password"
  ai-apm-ingest:
    environment:
      DORIS_USER: "root"
      DORIS_PASSWORD: "your-strong-password"
EOF

docker compose up -d ai-apm-web ai-apm-ingest
docker exec ai-apm-web printenv | grep '^DORIS_USER\|^DORIS_PASSWORD'
docker exec ai-apm-ingest printenv | grep '^DORIS_USER\|^DORIS_PASSWORD'

If the same override already has other settings, append under the matching services entries instead of replacing the whole file.

Kubernetes

bash

kubectl -n databuff edit configmap ai-apm-config

Under data::

yaml

  DORIS_USER: "root"
  DORIS_PASSWORD: "your-strong-password"

Then restart Web and Ingest:

bash

kubectl -n databuff rollout restart deploy/ai-apm-web deploy/ai-apm-ingest
kubectl -n databuff rollout status deploy/ai-apm-web
kubectl -n databuff rollout status deploy/ai-apm-ingest
kubectl -n databuff exec deploy/ai-apm-web -- printenv | grep '^DORIS_USER\|^DORIS_PASSWORD'

Same upgrade caveat as login credentials: re-applying the stock ConfigMap drops custom keys. The password lives in env vars—protect host access accordingly.

3. Configure Doris FE / BE addresses

Supports comma-separated multi-FE / multi-BE (host or host:port lists + health round-robin). Single-node defaults for one-line installs are unchanged.

What each port is for

Port (default)	Environment variable	Used by	Purpose
9030	`DORIS_FE_QUERY_PORT`	Web, Ingest	FE JDBC (MySQL protocol reads)
8030	`DORIS_FE_HTTP_PORT`	Mainly Ingest	FE HTTP; used for Stream Load only when `DORIS_BE_HTTP_HOST` is unset — PUT FE `_stream_load`, then 307 to BE
8040	`DORIS_BE_HTTP_PORT`	Ingest only	BE HTTP; when `DORIS_BE_HTTP_HOST` is set, Stream Load goes directly to BE (recommended)

One-line installs set DORIS_BE_HTTP_HOST by default, so writes use BE:8040 and 8030 is rarely used. Web queries use 9030 only — no Stream Load, no BE setting.

text

Query:  Web/Ingest  --JDBC-->  FE:9030
Write:  Ingest  --Stream Load-->  BE:8040           (BE configured; default)
   or:  Ingest  --Stream Load-->  FE:8030 -307-> BE (BE unset)

Parameters

Purpose	Component	YAML key	Environment variable	Default
FE host (multi OK)	Web / Ingest	`apm.doris.fe-host` / `ingest.doris.fe-host`	`DORIS_FE_HOST`	`127.0.0.1`
FE query port (JDBC)	Web / Ingest	`*.fe-query-port`	`DORIS_FE_QUERY_PORT`	`9030`
FE HTTP port (only when Stream Load goes via FE, i.e. BE unset)	Ingest (mainly)	`*.fe-http-port`	`DORIS_FE_HTTP_PORT`	`8030`
BE HTTP host (multi OK, Stream Load)	Ingest only	`ingest.doris.be-http-host`	`DORIS_BE_HTTP_HOST`	empty (see below)
BE HTTP port (default when host has no port)	Ingest only	`ingest.doris.be-http-port`	`DORIS_BE_HTTP_PORT`	`8040`

One-line Docker / K8s defaults:

Environment variable	Docker Compose default	Notes
`DORIS_FE_HOST`	`ai-apm-doris-fe`	Host for JDBC and (optional) FE HTTP
`DORIS_BE_HTTP_HOST`	`ai-apm-doris-be`	Ingest direct Stream Load to BE

Address and port format

Both FE and BE *_HOST values support these three forms (comma-separated; host-only and host:port may be mixed):

Form	FE example	BE example
Single host	`DORIS_FE_HOST=fe1`	`DORIS_BE_HTTP_HOST=be1`
Multi-host (shared port)	`DORIS_FE_HOST=fe1,fe2`	`DORIS_BE_HTTP_HOST=be1,be2`
Multi-host (per-entry port)	`DORIS_FE_HOST=fe1:19030,fe2:19031`	`DORIS_BE_HTTP_HOST=be1:8040,be2:8041`

When an entry omits the port, use the matching PORT env var above; if unset, defaults are 9030 / 8030 / 8040.

Notes:

fe1,fe2: JDBC uses 9030 (query); FE Stream Load (if used) uses 8030 (http) — each default applies on its path.
fe1:p1,fe2:p2: the explicit port applies to both JDBC and FE HTTP for that entry (one host field, one port). If query and http ports differ, prefer fe1,fe2 + separate PORT vars.
be1,be2 + DORIS_BE_HTTP_PORT=8040 matches the pre-multi-node host/port split; be1:8040,be2:8040 also works.
Multi-FE JDBC uses jdbc:mysql:loadbalance://....

With vs without BE host

Scenario	Behavior
`DORIS_BE_HTTP_HOST` set (recommended; Docker default)	Stream Load goes directly to BE:8040; multi-BE health round-robin, failover, `/api/health` probes. Does not use FE:8030
Unset (empty)	Stream Load hits FE:8030 `_stream_load`, then 307 to BE; less reliable with split FE/BE — keep BE set for one-line installs

Multi-host guidance

Writes (recommended): DORIS_BE_HTTP_HOST=be1,be2,be3 + DORIS_BE_HTTP_PORT=8040 (or be1:8040,be2:8040).
Queries: DORIS_FE_HOST=fe1,fe2 + DORIS_FE_QUERY_PORT; you usually do not need to change DORIS_FE_HTTP_PORT for writes.
Table replicas (replication_num ≥ 2) are cluster-side redundancy and complement the app address list.

Docker example

For an external Doris or custom addresses, use docker-compose.override.yml:

bash

cd /opt/databuff-ai-apm

cat > docker-compose.override.yml <<'EOF'
services:
  ai-apm-web:
    environment:
      DORIS_FE_HOST: "192.168.1.10,192.168.1.11"
      DORIS_FE_QUERY_PORT: "9030"
      DORIS_FE_HTTP_PORT: "8030"
  ai-apm-ingest:
    environment:
      DORIS_FE_HOST: "192.168.1.10,192.168.1.11"
      DORIS_FE_QUERY_PORT: "9030"
      DORIS_FE_HTTP_PORT: "8030"
      # Compatible form: hosts + shared port (same as pre-multi-node)
      DORIS_BE_HTTP_HOST: "192.168.1.20,192.168.1.21"
      DORIS_BE_HTTP_PORT: "8040"
      # Or ultra-style: DORIS_BE_HTTP_HOST: "192.168.1.20:8040,192.168.1.21:8040"
EOF

docker compose up -d ai-apm-web ai-apm-ingest

Kubernetes

bash

kubectl -n databuff edit configmap ai-apm-config

Under data:, for example:

yaml

  DORIS_FE_HOST: "192.168.1.10,192.168.1.11"
  DORIS_FE_QUERY_PORT: "9030"
  DORIS_FE_HTTP_PORT: "8030"
  DORIS_BE_HTTP_HOST: "192.168.1.20,192.168.1.21"
  DORIS_BE_HTTP_PORT: "8040"

Then restart Web and Ingest:

bash

kubectl -n databuff rollout restart deploy/ai-apm-web deploy/ai-apm-ingest

4. Adjust Ingest task parameters

Ingest uses worker pools and ring buffers for Trace / Metric / aggregation. Defaults live in ai-apm-ingest application.yml and can be overridden by env vars.

Common task and buffer settings

YAML key	Environment variable	Default	Meaning
`ingest.pipeline.trace-tasks`	`INGEST_TRACE_TASKS`	`8`	Trace parse/assemble parallelism
`ingest.pipeline.metric-tasks`	`INGEST_METRIC_TASKS`	`4`	Metric routing parallelism
`ingest.pipeline.aggregate-tasks`	`INGEST_AGGREGATE_TASKS`	`4`	Minute aggregation workers
`ingest.pipeline.trace-buffer-size`	`INGEST_TRACE_BUFFER_SIZE`	`8192`	Ring slots per trace worker (≥16)
`ingest.pipeline.metric-buffer-size`	`INGEST_METRIC_BUFFER_SIZE`	`1024`	Ring slots per metric worker
`ingest.pipeline.aggregate-buffer-size`	`INGEST_AGGREGATE_BUFFER_SIZE`	`1024`	Ring slots per aggregate worker

When buffers fill, events are dropped (overflow). If the UI misses data but ingest logs look healthy, raise *_BUFFER_SIZE or *_TASKS first. For Doris flush and related knobs, see Performance Tuning — Ingest pipeline.

Docker

Prefer docker-compose.override.yml:

bash

cd /opt/databuff-ai-apm

cat > docker-compose.override.yml <<'EOF'
services:
  ai-apm-ingest:
    environment:
      INGEST_TRACE_TASKS: "16"
      INGEST_TRACE_BUFFER_SIZE: "16384"
      INGEST_METRIC_TASKS: "8"
EOF

docker compose up -d ai-apm-ingest
docker exec ai-apm-ingest printenv | grep '^INGEST_'

If the same override already sets Web credentials, append under the matching services entries instead of replacing the whole file.

Kubernetes

bash

kubectl -n databuff edit configmap ai-apm-config

Under data::

yaml

  INGEST_TRACE_TASKS: "16"
  INGEST_TRACE_BUFFER_SIZE: "16384"
  INGEST_METRIC_TASKS: "8"

Restart ingest:

bash

kubectl -n databuff rollout restart deploy/ai-apm-ingest
kubectl -n databuff rollout status deploy/ai-apm-ingest
kubectl -n databuff exec deploy/ai-apm-ingest -- printenv | grep '^INGEST_'

Same upgrade caveat as login credentials: re-applying the stock ConfigMap drops custom keys.

Tuning tips

Change one or two knobs at a time so you can roll back easily.
Fix drops first (larger buffers / tasks), then tune against CPU and Doris Stream Load pressure.
Higher *_TASKS uses more CPU and memory—see the baselines in Performance Tuning.

5. Trace resource ignore filtering

Drop matching spans (skip enrich / assemble / write and metrics) to filter noise such as health checks, Prometheus scrapes, and SELECT 1. Matching uses the span resource and, for HTTP, metaHttpUrl (either hit drops the span). Exact rules use full-string equality; regex uses Java Matcher#matches() (full-string match).

YAML key	Environment variable	Default	Meaning
`ingest.trace.ignore-resources`	`INGEST_TRACE_IGNORE_RESOURCES`	empty	Exact-match list; comma-separated in env
`ingest.trace.ignore-resource-regex`	`INGEST_TRACE_IGNORE_RESOURCE_REGEX`	empty	Full-string regex list; comma-separated in env

Option 1: Edit `application.yml`

For local development or when you mount / bake config into the image. Edit ingest application.yml:

yaml

ingest:
  trace:
    ignore-resources:
      - PING
      - /actuator/prometheus
    ignore-resource-regex:
      - ^/actuator(/.*)?$
      - ^SELECT 1$

Option 2: Docker Compose environment variables

Preferred for one-line installs — use docker-compose.override.yml (survives upgrades):

bash

cd /opt/databuff-ai-apm   # or: echo $APM_INSTALL_DIR

cat > docker-compose.override.yml <<'EOF'
services:
  ai-apm-ingest:
    environment:
      INGEST_TRACE_IGNORE_RESOURCES: "PING,/actuator/prometheus"
      INGEST_TRACE_IGNORE_RESOURCE_REGEX: "^/actuator(/.*)?$,^SELECT 1$"
EOF

docker compose up -d ai-apm-ingest

If the same override already sets other ingest knobs, append under environment instead of replacing the whole file. If a regex or path contains commas, prefer Option 1 (YAML lists) to avoid delimiter ambiguity.

Kubernetes

bash

kubectl -n databuff edit configmap ai-apm-config

Under data::

yaml

  INGEST_TRACE_IGNORE_RESOURCES: "PING,/actuator/prometheus"
  INGEST_TRACE_IGNORE_RESOURCE_REGEX: "^/actuator(/.*)?$,^SELECT 1$"

Restart ingest:

bash

kubectl -n databuff rollout restart deploy/ai-apm-ingest
kubectl -n databuff rollout status deploy/ai-apm-ingest

Verify

After restart, ingest logs should contain Span resource ignore filter enabled when rules are loaded.

6. SkyWalking SQL normalization

Applies only to SkyWalking ingest (raw SQL from the agent). OTLP / DataBuff agents already normalize on the client; ingest does not touch them. Default mode=1 matches open-source before v0.1.7; override via env. When enabled (not -1), replaced literals are also written to OTel db.query.parameter.0, 1, 2, … (SQL only; HTTP unchanged).

YAML key	Environment variable	Default	Meaning
`ingest.skywalking.sql-normalized-type`	`INGEST_SQL_NORMALIZED_TYPE`	`1`	`-1` disables normalize and parameter extraction; `0` digit-prefix → `?`; `1` contains digit → `?`

Docker

bash

cd /opt/databuff-ai-apm   # or: echo $APM_INSTALL_DIR

# Append under existing override ai-apm-ingest.environment, e.g. to disable:
#   INGEST_SQL_NORMALIZED_TYPE: "-1"

docker compose up -d ai-apm-ingest
docker exec ai-apm-ingest printenv | grep '^INGEST_SQL_NORMALIZED_TYPE'

Startup log SkyWalking SQL normalize mode=-1 (or 0/1) means the setting is active.

Kubernetes

Add INGEST_SQL_NORMALIZED_TYPE: "-1" under ConfigMap data:, then:

bash

kubectl -n databuff rollout restart deploy/ai-apm-ingest

7. Adjust storage retention

To keep data for 14 days, connect to Doris and run the following SQL (no FE / BE restart required):

sql

USE databuff;
ALTER TABLE trace_dc_span SET ("dynamic_partition.start" = "-14");
ALTER TABLE log_dc_record SET ("dynamic_partition.start" = "-14");
ALTER TABLE metric_service SET ("dynamic_partition.start" = "-14");
-- Repeat for other metric_* tables as needed

8. Skip AVX2 CPU check (install time)

Doris BE on x86_64 / amd64 requires AVX2. Online/offline installers check the CPU and exit 1 if the avx2 flag is missing. For PoC or legacy VMs only, export this env var before install:

Environment variable	Value	Meaning
`DATABUFF_SKIP_AVX2_CHECK`	`1` or `true`	Skip the installer's AVX2 check (warning only, does not abort)

Online install:

bash

export DATABUFF_SKIP_AVX2_CHECK=1
curl -fsSL https://databuff.ai/databuff/ai-apm-install.sh | bash

Offline install:

bash

export DATABUFF_SKIP_AVX2_CHECK=1
cd /path/to/databuff-docker-offline-*-amd64
./install-offline.sh

This only bypasses the installer gate; it does not remove Doris's AVX2 dependency. Prefer AVX2-capable x86_64 or arm64 for production. Full risks and legacy-script workarounds: Performance Tuning — Bypass AVX2 check.

Parameter Configuration ​

1. Change login username and password ​

Docker ​

Kubernetes ​

2. Configure Doris username and password ​

Docker ​

Kubernetes ​

3. Configure Doris FE / BE addresses ​

What each port is for ​

Parameters ​

Address and port format ​

With vs without BE host ​

Multi-host guidance ​

Docker example ​

Kubernetes ​

4. Adjust Ingest task parameters ​

Common task and buffer settings ​

Docker ​

Kubernetes ​

Tuning tips ​

5. Trace resource ignore filtering ​

Option 1: Edit application.yml ​

Option 2: Docker Compose environment variables ​

Kubernetes ​

Verify ​

6. SkyWalking SQL normalization ​

Docker ​

Kubernetes ​

7. Adjust storage retention ​

8. Skip AVX2 CPU check (install time) ​

Related docs ​

Parameter Configuration

1. Change login username and password

Docker

Kubernetes

2. Configure Doris username and password

Docker

Kubernetes

3. Configure Doris FE / BE addresses

What each port is for

Parameters

Address and port format

With vs without BE host

Multi-host guidance

Docker example

Kubernetes

4. Adjust Ingest task parameters

Common task and buffer settings

Docker

Kubernetes

Tuning tips

5. Trace resource ignore filtering

Option 1: Edit `application.yml`

Option 2: Docker Compose environment variables

Kubernetes

Verify

6. SkyWalking SQL normalization

Docker

Kubernetes

7. Adjust storage retention

8. Skip AVX2 CPU check (install time)

Related docs