Closed Stephan3555 closed 5 years ago
Sounds like an AWS error, not really a druid one, no?
Does it retry to download the file?
Hi patelh, thanks for your quick response. Could be an AWS error, thats the reason why I posted here. So maybe others experience this as well. Regarding your second question, after a while the download/upload to s3 seems to work. So it seems to be only temporarely unavailable.
We are not sure what the reason behind this is. Could be a networkproblem with our Rancher installation as well. Or the jets3t Library is not fully compatible with AWS/S3 anymore (Jets3t-0.9.4 is from August 2015 and since then never updated)
But its strange that it started a few days ago till now. Overall the Druid-Cluster is working but the Historical logs got flooded with this error messages
We found the cause for our problem. For us the problem was the amount of segments we had in our cluster. We had around 21k segments distributed on three historicals. After some debugging we found out that the problem was the connection pool size:
We merged the incredible small segments to larger segments and now the error disappeared.
Hi,
since a few days our Historical Nodes on AWS have temporarily problems connecting to S3. The Data are still written in S3, but from time to time follwoing error occurs (I replaced sensitive Information with <>):
ERROR-Message:
We are using currently Druid 0.12.3 with following configuration:
common.runtime.properties:
jets3t.properties:
Does anybody else experience this behavior? I gladly provide more information/configuration to solve this issue.
Thanks, Stephan