Open runningman84 opened 5 years ago
I have my Graylog cluster setup using DocumentDB and everything else works other than that one collection. I'm still testing some ideas for keeping it at a reasonable size. Also 🤞that DocumentDB will support capped collections sometime soon.
I migrated an existing graylog setup from mongo to documentdb by using AWS DMS ... it is working fine so far, all collections were created on documentdb but they lost the capped capability is there any disadvantage from not recycling these collections?
@benishak Since the collection isn't capped it can get to a point where when Graylog tries to scan it, all of the read credits will get used up. You'll have pretty bad time once that happens.
Isn't there an option to disable the whole dead letter feature?
from source code index failures logs is the one using a capped collection and limiting logs to 50MB these logs or this collection is not being actively scanned or read as I understood. I believe if this collection grows significantly very fast then there is an issue in the cluster.
That's exactly what happened to me, I had an issue where 1,000s of records were being written to the index failure table. I never figure which part of the system was trying to read that table but it caused major issues at that point.
I'm running 2.5 still so I don't know if anything is changed in 3.x.
Came across this issue today deploying onto AWS DocDB.
2020-05-04 00:52:37,036 ERROR: org.graylog2.bootstrap.CmdLineTool - Guice error (more detail on log level debug): Error injecting constructor, com.mongodb.MongoCommandException: Command failed with error 303: 'Feature not supported: capped:true' on server XXX:27017. The full response is
{
"ok": 0,
"errmsg": "Feature not supported: capped:true",
"code": 303
}
From AWS what is supported and what not.
Any work around this issue?
Have you opened a feature request with AWS? They don't support capped capability. I think that would be the only thing stopping this from working, unfortunately. :(
There's an open feature request with AWS. I recommend adding your name to it via your TAMs.
Expected Behavior
Graylog should work with AWS DocumentDB which is very similar to mongodb.
Current Behavior
Graylog does not work with DocumentDB because it uses a function "capped" which is not supported by DocumentDB.
Possible Solution
Maybe this function is not really needed and the same feature could be implemented in a way which does not need this function.
Steps to Reproduce (for bugs)
Context
This feature would allow users to run Graylog in AWS Fargate and use AWS DocumentDB and AWS Elasticsearch for the heavy lifting.
Your Environment