-
-
Notifications
You must be signed in to change notification settings - Fork 196
Another " java.lang.IllegalArgumentException: Invalid negative offset" #120
Comments
I've also run into this on a Kafka 2.3.1 cluster; still investigating but this is not specific to kafka-lag-exporter. Loop over all the consumer groups and simply pass those into the
In my case, when I try to list the members of this group, the tool reports that there are no members.
After deleting this group, kafka-lag-exporter stops crashing.
|
@dylanmei Thanks for checking. I can't recreate this issue so I don't have much to go on. The exception itself is exactly what it sounds like, an offset returned by a Do you have any theories here? Next time it happens could you describe the topics involved and paste their configuration here? |
I will check the groups as @dylanmei suggested and return. |
I am having a hard time in order to get those scripts running over Azure EventHub for Kafka... |
I am unfamiliar with Azure EventHub. However, you can pass a bootstrap.servers={YOUR.EVENTHUBS.FQDN}:9093
security.protocol=SASL_SSL
sasl.mechanism=PLAIN
sasl.jaas.config=org.apache.kafka.common.security.plain.PlainLoginModule required username="$ConnectionString" password="{YOUR.EVENTHUBS.CONNECTION.STRING}"; Then run:
|
@dylanmei , the MS support gave us an important information about Azure EventHub For Kafka: new consumer group has an offset value of -1. I don´t know if Kafka itself works like this too, but that info and your hint about We found indeed topics without any message produced but with consumer groups created for it anyway, which then caused the The current workaround was to insert a dummy message so the offset was set. Thank you very much |
@andreminelli That's an interesting discovery. Thanks for following up. I suppose you could also use Kafka Lag Exporter whitelisting to ignore groups that have offsets initialized this way, as long as you know what they are ahead of time. |
I'm going to leave this open for now. Worst case we can catch and handle this exception message and report |
This handling makes sense - stop the whole topic checking because of some consumer groups 'sitting idle' is terrible... |
A fix for this is on its way in Kafka 2.4.1: https://issues.apache.org/jira/browse/KAFKA-9507 |
Ran into it also. Please consider the NaN by catching that exception as it will be at least year until that latest kafka version is upgraded in production... Can't use blacklisting groups workaround as group naming is under devs control.. |
I expect the fix to appear in the Kafka native client library used by this exporter. There will be no need to upgrade your clusters. |
Yes, the fixed admin client should resolve the issue. I'll cut a release next week. |
@seglo any updates on new release? |
Pretty please ;) |
0.6.1 is released with a Kafka clients bump to 2.5.0. I've had trouble reproducing this issue so it would be great if somebody in this thread could verify this is resolved. Thanks! |
Hi, I was a bit behind on my kafka-lag-exporter maintenance and was running the container To find out where the regression occurred, I upgraded sequentially and had no issues with I did loop through our kafka-consumer-groups:
I found a bunch that had no active members:
We are running confluent-kafka 6.2.10. I hope this is helpful. Looking through the diff I see this is the first time the kafka client has been upgraded, from 2.5.0 to 3.2.1. So the regression must be upstream. |
I am getting the following error - the message is the same as issue 41, but already running with 0.5.5 (and just tried 0.6.0, with same problem):
This is trying to connect to Azure Event Hub for Apache Kafka. Some instances work, others fail with this problem.
Any thoughts?
Regards.
Any thoughts?
The text was updated successfully, but these errors were encountered: