Relay troubleshooting tips (v2.X.X)

Introduction

Devo Relay 2.X.X is based on a Java process that runs on a Linux machine. The Java process internally uses a syslog engine called Scoja.

The relay Java process runs as a service in the Linux OS and it is managed by the system manager systemd. Once running, the relay listens for TCP/UDP connections on ports and, depending on the port, modifies the syslog events and forwards them to the Devo central collector / ELB. The main role of the relay is to tag events based on a defined rule, then forward the tagged events to Devo. 

The relay comes with a few ports that are preconfigured with rules for managing specific types of events. For example, port 13000 is preconfigustartured to forward the events received directly to Devo without any other processing. This port should be used for events that are properly tagged in the data source. You can also create custom rules to tag and/or filter the events sent to the associated port on the relay. Each rule is stored in one of these two folders:

• /opt/devo/ng-relay/conf/relay/run/rules/ → Rules managed from the Devo web application.

• /opt/devo/ng-relay/conf/relay/run/unrules/ → Rules not managed from the Devo web application. These rules are directly managed on the relay machine itself.

About the relay logs files

Main process logs

As the relay runs as a systemd service, the logs can be consulted using the following command:

These log files are also sent to Devo by default and can be consulted in the table box.unix.

Scoja logs

In addition, several Scoja log files are generated and stored in:

/var/logt/local/<year>/<month>/<day>/<relayName>/syslog/scoja/

• main.log → main process log file

• source.log → logs related to the sources configured

• thread.log → threads opened for each source

• stats.log → statistics about received and sent events

• target.log  → logs related to the sending process

These logs are also sent to Devo and can be consulted in the tables syslog.scoja.*. However, we recommend keeping them in the local machine for a few days.

The retention configuration can be set by updating the following file:

/opt/devo/ng-relay/conf/relay/maduro.properties

Troubleshooting checklist

Make sure the relay process (Java) is running 

Response

Check the relay process logs and check for errors

To actively follow the logs as they are being written, you can use the -f flag. This works as you might expect if you have experience using tail -f.

Note that you will see several exceptions (Connection refused) until Scoja starts. That’s normal behavior. In order to avoid some redundancy in the logs, you can use the following command:

Response

Confirm that the default ports are listening

• Port 12996 TCP where an HTTP server is listening.

• Port 12997 TCP where scoja will be listening for local syslog events.

• Port 12998 TCP where the relay will be listening for commands from the Devo Relay CLI.

• Port 12999 is listening on UDP and is prepared to receive external Netflow traffic (Netflow version < 9). 

• Ports 13000-13002 are listening on both TCP and UDP.

• Port 5140 is listening on TCP and UDP to handle internal relay logs. 

Response

Ensure that the client certificate is installed

The client certificate file (client.jks) should be installed in the relay's /opt/devo/ng-relay/conf/relay/run/keys folder. 

Response

If the certificate download fails, you can force the download again:

1. Log in to the Devo web application.

2. Go to Administration → Relays and ELBs → Relays, click the ellipsis menu of the required relay, and select Edit. The Relay Input (Rules) screen appears. 

3. Select the Force Generate New Certificate at the bottom of the screen, then click the Apply Configuration button.

Ensure that scoja has started and relay configuration is being updated correctly

A cron task runs every minute to check if any changes to the relay configuration have been made in the Devo web application. The result of this check appears in the journal of the service.

Response

• The "[SUCCESS] Relay activation is still pending, waiting for user to activate the Relay from the web app" message indicates that the relay is registered but has not yet been activated in the Devo web application.

• The "[SUCCESS] Downloading and deploying configuration set" message confirms that the relay has been activated and a new configuration has been downloaded.

Once activated, you will be able to define the relay rules using the web application. The relay will begin checking the configuration for updates every 1 minute.

• The "[SUCCESS] Relay is active. No pending configuration changes to be applied" message occurs every minute to confirm that no changes were detected in the relay configuration.

During the first configuration, the certificate is also downloaded. If the certificate download fails, you must force the download again.

Error when not enough memory available

Error when no connection with the collector / ELB

Send test events to confirm the relay is receiving and forwarding correctly

Test through the CLI

In order to test that the relay works normally, you can use the CLI command test that will send several events to the relay.

Response

You can also check that the events are in the table syslog.relay.monitor.

Test with echo and netcat

Alternatively, echo and netcat commands can be used to send several events to the relay port 13000 to confirm that it is receiving and forwarding events to Devo.

Enter this command on the relay machine to send 100 test events to test.keep.free table:

To confirm that the events were processed correctly, open Devo, go to Data Search, and locate the test.keep.free table. This table should contain 100 events. 

If there are no events logged in the test.keep.free table, check the unknown.unknown table. If there was a problem applying the tags, events will be logged in this table.

Check if the firewall is enabled for the ports

Use the following commands to enable the required sending ports:

Copy

Some common problems and solutions

Check relay rules with regexes are efficient

We have updated the relay rules related to several technologies with a more efficient regex.

• Paloalto rule - we need to extract the capture group 1 but we don’t need to process the whole message.

  • Inefficient regex:  ^[^,]+,[^,]+,[^,]+,([^,]+).*$

  • Efficient regex: ^[^,]+,[^,]+,[^,]+,([^,]+)

Rule is misconfigured and the capture group is not pointing to the correct regex field

If you see the following message then check here for details:

Issues pertaining to Java memory error logs

In case there are java.lang.OutOfMemoryError errors in the logs inside syslog.scoja.target table like in the example below.

Try increasing the memory on the Relay. In certain cases increasing it to 6GB of RAM has proven to work.

Relay runs into IndexOutOfBoundsException

When sending events through a relay, it’s possible to run into an IndexOutOfBoundsException if the regex (regular expression) in the relay rule you specified doesn’t conform to the messages that you’re sending to the relay. For example, if you have the event bsssbssssbsss and you have a regex rule that says locate the third b and return everything until the end of the message, the result would be sss. However, if you sent the event bsssbssss using the same regex rule, you would get the IndexOutOfBoundsException because you're trying to access a character that doesn't exist and goes beyond what's contained in the event.

1. Analyze and verify that the error message indicates an out of bounds exception related to the regex rule.

2. Analyze the messages that are being sent through the relay and compare them to the regex rule you have defined in the relay to see what inconsistencies there are in terms of the format.

3. Modify the regex rule in the relay so that it conforms to the format of the events that you’re sending through the relay.

Relay rules not applied

Relay rules don’t seem to be applied on the relay after attempting to add a rule within Devo's web GUI.

1. Check the table for events showing that a new configuration was downloaded and deployed.

2. Validate rule added to relay via CLI. Check the syslog.relay.conf table to confirm the rule has been applied to the relay.

3. If neither shows up then double-check the relay's connectivity.

In the event the buffer becomes full

The Devo Relay includes a disk buffering mechanism to store events in case of connectivity loss. Once connectivity is restored, the logs stored in the disk buffer are sent to the collector.

However, if the connectivity issue persists for too long, the disk buffer may become full, requiring manual intervention to recover the instance.

The proposed steps for the administrator are the following:

1. Verify the location of the disk buffer folder, typically `/var/log/relay/buffer`. Check the relay configuration in case the folder has been configured to a different location.

2. Ensure the mount point associated with the disk buffer is not full. Use the `df` command to check disk usage.

3. If the mount point is full

   1. Stop the relay service: systemctl stop devo-ng-relay.

   2. Identify the .sdb files located in /var/log/relay/buffer. These files store the disk buffer data. Perform one of the following actions:

      1. Relocate .sdb files to another partition to free up space. Move them back later to ensure the data is sent and not lost.

      2. Transfer .sdb files to another relay with the same relay version to distribute the load of sending events.

      3. Remove sdb only if data loss is acceptable.

   3. Manually start the service again and verify that the syslog sender begins shipping logs to the platform: systemctl start devo-ng-relay .

   4. Verify the relay is working properly with devo-ng-relay-cli console tool and its test command.

   5. Note that uploading .sdb files may take a long time depending on the amount of stored data. You can monitor the process in the syslog.scoja.stats table with fields totalBufferLogsSecondary and partialBufferLogsSecondary.

   6. .sdb files in /var/log/relay/buffer are not removed or truncated during the upload process. They are only removed after being fully uploaded. You can determine the currently uploading .sdb file by checking its timestamp (updated while being processed).