Skip to main content
logoTetrate Service BridgeVersion: 1.4.x

Configure Log Levels

This document describes how to adjust log levels for the different components in TSB, including platform components, Envoy sidecars and ingress gateways at runtime, as well as the procedure to view the logs.

Before you get started make sure:

✓ You have installed and configured TSB properly.
✓ You have installed and configured kubectl to access the application cluster.

For the example commands we assume that you have some applications deployed in a helloworld namespace.

TSB Components Produce Lots of Logs

Be careful enabling debug logging across all of TSB's scopes for extended periods of time - TSB components produces a lot of logs! You may be faced with large log ingestion bills due to automatic log ingestion combined with turning up log levels across TSB or Sidecars.

List the available components

In order to change each components' log level you will need to know which components are available. For that, there is utility command in tctl that will leverage the current kubectl connection information (context) and list the available components in that cluster.

tctl experimental debug list-components 
Output
PLANE         COMPONENT                 DEPLOYMENTS                           
management ldap ldap
management mpc mpc
management xcp central, xcp-operator-central
management frontenvoy envoy
management iamserver iam
management apiserver tsb
management tsb-operator tsb-operator-management-plane
management web-ui web
management zipkin zipkin
management oap oap
management collector otel-collector
management postgres postgres
control istio istio-operator
control hpaadapter istio-system-custom-metrics-apiserver
control oap oap-deployment
control onboarding onboarding-operator
control collector otel-collector
control tsb-operator tsb-operator-control-plane
control zipkin zipkin
control xcp edge, xcp-operator-edge
data operator istio-operator
data tsb-operator tsb-operator-data-plane
data bookinfo-gateway bookinfo/bookinfo-gateway
data helloworld-tls-gateway helloworld-tls/helloworld-tls-gateway
data helloworld-gateway helloworld/helloworld-gateway
data httpbin-gateway httpbin/httpbin-gateway
data tier1 tier1/tier1

As seen in the output above, this command will list all available components in the cluster and sort them by plane (management, control or data plane). It will also show the Kubernetes deployments that build up every component. The PLANE and COMPONENT columns in the output is what will need to be used with the command to set the log level below. For instance, to change the mpc component log level, you will need to refer to it with management/mpc.

TSB platform components (management and control planes)

TSB components are able to adjust the log levels for the different existing loggers at runtime without restarting the pod. For that, a new command in tctl CLI has been added.

In order to check the available loggers for a component and check the current levels, run the command without any flag.

tctl experimental debug log-level management/iamserver
Output
Configuring the logging levels:
POST /logging?level=value -> Configures all levels globally
POST /logging?logger=value -> Configures the logging level for 'logger'

Current logging levels:

admin info Administration server logs
auth info Authentication messages server
config info Messages from the config system
credentials/basic info Credentials parsing provider for basic http
credentials/jwt info Credentials parsing provider for JWT bearer
default info Unscoped logging messages.
dynadsn info Messages from dynamic db conn pool
envoy-filter info Envoy filter messages
exchange info Messages from token exchange
grpc info Messages from the gRPC layer
health info Messages from health check service
iam-server info Messages from the RunGroup handler
iam/http info Messages from http-server
jwt info Messages from the LDAP provider
keyvalue/tx info Messages from the transaction system
ldap info LDAP integration messages
local info Messages from the local authentication provider
migrations info Database migration messages
oauth info Messages from the Server Extensions
oauth2 info OAuth2 messages
oidc info Messages from the OIDC provider
root info Messages from the root credentials package
server info Messages from service main

In the output above, the leftmost column shows the logger name, the middle column shows the current log level configured for that given logger, and the last column shows a brief description of the kind of messages that logger shows.

In order to change the log levels, there are multiple ways to accomplish that, with different combinations of the level flag.

Change a single logger

Changing a single logger is possible by providing a logger name followed by a colon (:), followed by the desired level. For example:

tctl experimental debug log-level management/iamserver --level ldap:debug
Output
Configuring the logging levels:
POST /logging?level=value -> Configures all levels globally
POST /logging?logger=value -> Configures the logging level for 'logger'

Current logging levels:

admin info Administration server logs
auth info Authentication messages server
config info Messages from the config system
credentials/basic info Credentials parsing provider for basic http
credentials/jwt info Credentials parsing provider for JWT bearer
default info Unscoped logging messages.
dynadsn info Messages from dynamic db conn pool
envoy-filter info Envoy filter messages
exchange info Messages from token exchange
grpc info Messages from the gRPC layer
health info Messages from health check service
iam-server info Messages from the RunGroup handler
iam/http info Messages from http-server
jwt info Messages from the LDAP provider
keyvalue/tx info Messages from the transaction system
ldap debug LDAP integration messages
local info Messages from the local authentication provider
migrations info Database migration messages
oauth info Messages from the Server Extensions
oauth2 info OAuth2 messages
oidc info Messages from the OIDC provider
root info Messages from the root credentials package
server info Messages from service main

You can see by the output received that the ldap logger has changed its level to debug, raising its verbosity level.

Change multiple loggers

Changing multiple loggers at once is possible by providing a comma (,) separated list of logger name and level pairs. Items within a pair are separated by a colon (:). For example:

tctl experimental debug log-level management/iamserver --level jwt:error,auth:warn,health:error
Output
Configuring the logging levels:
POST /logging?level=value -> Configures all levels globally
POST /logging?logger=value -> Configures the logging level for 'logger'

Current logging levels:

admin info Administration server logs
auth warn Authentication messages server
config info Messages from the config system
credentials/basic info Credentials parsing provider for basic http
credentials/jwt info Credentials parsing provider for JWT bearer
default info Unscoped logging messages.
dynadsn info Messages from dynamic db conn pool
envoy-filter info Envoy filter messages
exchange info Messages from token exchange
grpc info Messages from the gRPC layer
health error Messages from health check service
iam-server info Messages from the RunGroup handler
iam/http info Messages from http-server
jwt error Messages from the LDAP provider
keyvalue/tx info Messages from the transaction system
ldap debug LDAP integration messages
local info Messages from the local authentication provider
migrations info Database migration messages
oauth info Messages from the Server Extensions
oauth2 info OAuth2 messages
oidc info Messages from the OIDC provider
root info Messages from the root credentials package
server info Messages from service main

You can see how only the selected loggers have changed to the specified levels.

Change all loggers at once

You can also change all loggers at once to a given level by just providing the level name, for instance:

tctl experimental debug log-level management/iamserver --level info
Output
Configuring the logging levels:
POST /logging?level=value -> Configures all levels globally
POST /logging?logger=value -> Configures the logging level for 'logger'

Current logging levels:

admin info Administration server logs
auth info Authentication messages server
config info Messages from the config system
credentials/basic info Credentials parsing provider for basic http
credentials/jwt info Credentials parsing provider for JWT bearer
default info Unscoped logging messages.
dynadsn info Messages from dynamic db conn pool
envoy-filter info Envoy filter messages
exchange info Messages from token exchange
grpc info Messages from the gRPC layer
health info Messages from health check service
iam-server info Messages from the RunGroup handler
iam/http info Messages from http-server
jwt info Messages from the LDAP provider
keyvalue/tx info Messages from the transaction system
ldap info LDAP integration messages
local info Messages from the local authentication provider
migrations info Database migration messages
oauth info Messages from the Server Extensions
oauth2 info OAuth2 messages
oidc info Messages from the OIDC provider
root info Messages from the root credentials package
server info Messages from service main

All of the loggers have been changed to the info level with a single command.

Configure log levels for ingress gateways

In order to change the gateways log levels, the same procedure described above can be used. Note that the list-components command output also includes the gateways deployed in the current cluster under the data plane.

tctl experimental debug list-components  | egrep ^data
Output
data          tsb-operator              tsb-operator-data-plane                  
data operator istio-operator
data bookinfo-gateway bookinfo/bookinfo-gateway
data helloworld-tls-gateway helloworld-tls/helloworld-tls-gateway
data helloworld-gateway helloworld/helloworld-gateway
data httpbin-gateway httpbin/httpbin-gateway
data tier1 tier1/tier1

The procedure to change the log level for a gateway will be the same as for the rest of the components. For instance, to verify the bookinfo-gateway log levels, the following command can be run:

tctl experimental debug log-level data/bookinfo-gateway
Output
active loggers:
admin: trace
aws: trace
assert: trace
backtrace: trace
cache_filter: trace
client: trace
config: trace
connection: trace
conn_handler: trace
decompression: trace
dubbo: trace
envoy_bug: trace
ext_authz: trace
rocketmq: trace
file: trace
filter: trace
forward_proxy: trace
grpc: trace
hc: trace
health_checker: trace
http: trace
http2: trace
hystrix: trace
init: trace
io: trace
jwt: trace
kafka: trace
lua: trace
main: trace
matcher: trace
misc: trace
mongo: trace
quic: trace
quic_stream: trace
pool: trace
rbac: trace
redis: trace
router: trace
runtime: trace
stats: trace
secret: trace
tap: trace
testing: trace
thrift: trace
tracing: trace
upstream: trace
udp: trace
wasm: trace

And the log levels can be adjusted using the same procedure, for instance to turn all logger to info level, the following command can be used:

tctl experimental debug log-level data/bookinfo-gateway --level info
Output
active loggers:
admin: info
aws: info
assert: info
backtrace: info
cache_filter: info
client: info
config: info
connection: info
conn_handler: info
decompression: info
dubbo: info
envoy_bug: info
ext_authz: info
rocketmq: info
file: info
filter: info
forward_proxy: info
grpc: info
hc: info
health_checker: info
http: info
http2: info
hystrix: info
init: info
io: info
jwt: info
kafka: info
lua: info
main: info
matcher: info
misc: info
mongo: info
quic: info
quic_stream: info
pool: info
rbac: info
redis: info
router: info
runtime: info
stats: info
secret: info
tap: info
testing: info
thrift: info
tracing: info
upstream: info
udp: info
wasm: info

This will adjust the log levels to all replica pods of the gateway deployment.

Using istioctl to configure log levels of the data plane

Services that have been deployed in the service mesh can have their logging controlled dynamically. There are a few ways to change these levels, but the easiest is using the istioctl proxy-config log command.

istioctl proxy-config log <pod-name[.namespace]> --level <arguments>

The arguments can be in either of the following forms: level=<name> or <logger>=<name>.

When using the level=<name> form, all applicable components are set to the log level specified by name. When using the <logger>=<name> form, the log level on the particular logger specified by the logger is changed. Finally, you can list many loggers in a single command, like <logger1>=<name1>,<logger2>=<name2>,<logger3>=<name3>.

The following names are allowed: none, default, debug, info, warn, or error

For details on the different log levels and loggers available, please refer to the documentation of istioctl proxy-config log

You can view log levels directly with kubectl too!

Services that have been deployed in the service mesh will contain the pilot-agent command in the sidecar container. It's primary responsibility is to bootstrap the Envoy proxy, but it can also be used to configure logging levels at runtime among other things.

The pilot-agent command may be invoked to update the logging level through kubectl in the following way:

kubectl exec <pod-name> -c istio-proxy -- \
pilot-agent request POST 'logging?<arguments>'

pod-name refers to the target Kubernetes pod. Notice we are using -c istio-proxy option to explicitly specify that we are executing the pilot-agent command in the sidecar of the service that is deployed in pod-name

Configure log levels for application sidecars

Verify the deployed application pods

Verify that istio-proxy sidecars are properly deployed:

kubectl get pods -n helloworld -o jsonpath="{.items[*].spec.containers[*].name}" | \
tr -s '[[:space:]]' '\n' | \
sort | \
uniq -c | \
grep istio-proxy

This should print a text resembling the following output:

  2 istio-proxy

Adjust the log level

For this example we assume that the following applications have already been deployed and onboarded into TSB:

NAME                             READY   STATUS    RESTARTS   AGE
helloworld-v1-776f57d5f6-2h8dq 2/2 Running 0 5h49m
helloworld-v2-54df5f84b-v2wv6 2/2 Running 0 5h49m
istioctl proxy-config log helloworld-v1-776f57d5f6-2h8d --level debug
warning

It is recommended that you do NOT turn on debug log level for production workloads or workloads for high volume traffic systems. They may print excessive information that could overwhelm your application(s), or at the very least cost you a lot of money in log ingestion fees!

Once the above command takes effect, you will be able to view the debug logs from the sidecars using kubectl:

kubectl logs -f helloworld-v1-776f57d5f6-2h8dq -c istio-proxy

If you would like to apply the same change to the log level to other sidecars on your application, you will have to repeat the process for each pod that you are interested in.

Changing Log Levels with Kubectl

The istio-proxy sidecar in the application pod contains the pilot-agent command.

Run the following command via kubectl on the sidecar to configure the log level:

kubectl -n helloworld exec helloworld-v1-776f57d5f6-2h8dq -c istio-proxy -- \
pilot-agent request POST 'logging?level=debug'

Resetting the log levels

Once you are done inspecting the logs, always make sure to adjust the log level back again. This will also have to be done for each of the sidecars whose log levels you have adjusted.

istioctl has a shortcut for this:

istioctl proxy-config log helloworld-v1-776f57d5f6-2h8dq --reset
Reset shortcut not available for kubectl

Unfortunately there's not a shortcut for resetting the log levels via kubectl. You need to kubectl exec a command that restores all of the logs you changed using a list of log scopes and levels, like:

kubectl -n helloworld exec helloworld-v1-776f57d5f6-2h8dq -c istio-proxy -- \
pilot-agent request POST 'logging?h2=debug,http=info,grpc=error'