IllegalStateException with Message “unexpected message type: UnpooledUnsafeDirectByteBuf” from Netty #47
Comments
Hi @michaelahlers this is likely related to an account issue - we are investigating. |
@michaelahlers is this still happening? |
Closing this as it's not reproducible and may have been related to the temporary issue |
I got this issue recently in my project. I even migrated to newer datasift client version (from "3.0.0-Beta3.5" to "3.1.0") but this did not help. I am running my service on Amazon EC2 server. |
Hi @nbarabash - the original issue was related to an account specific problem. Changing the client version shouldn't make a difference, the connection reset error implies the connection is being severed. A few things could possibly cause this including proxies, firewall rules or an issue on the DS server side. Since we've had no other report of this I'm inclined to think the issue not be the latter. That said.... Do you have a proxy between the client and DataSift? What does your IP tables configuration looks like for outgoing connections on port 80 and 443? To narrow that down, have you configured the client to connect with or without SSL? SSL will connect to DS on port 443, without SSL it's port 80. |
Hi @zcourts, I don`t have any proxy between the client and DataSift. The issue appeared accidentally, worked fine a few weeks even with old client "3.0.0-Beta3.5". |
When did the issue first appear? Can you give me a stack trace of what is produced when you get the issue? |
@zcourts - this issue was spotted the first time on 2014-10-0. Data were lost for that period of time and I checked the logs, may be such exceptions were before that date, I am not sure. There is also another issue - client been disconnected from stream and does not connect again. There were not any changes in application related to DataSift stream connection. I`ll check EC2 firewall settings and will let you know. The stack trace is: |
That doesn't seem related to this issue. Opened #71 - I'll investigate Monday when I'm back in the office. |
We had this problem too, here is the stack trace : Feb 9 11:35:41 datasift-nexamaster io.netty.handler.codec.EncoderEx java.lang.IllegalStateException: unexpected message type: UnpooledUnsafeDirectByteBuf We usually have this problem after a couple of hours having our Datasift Streamer running. We run the same code over two identical VMs on independant environements. One has low amount of streams(3) using one connection, the other has a high amount of streams(40) with multiple connections(2-3). I took a look at the last 3 traces and (i'm probably wrong) I wonder if it could be related to multi-threading and the encoders (chunk vs initial state) ? |
Do you know if this happens while the client is attempting to reconnect? It's hard to say what could cause this other than an attempt to reconnect and there's a race condition which results in a message being written before the pipeline is configured or written at the end of the pipeline. Either of those would result in the Netty stack you've posted Does this consistently happen every couple of hours? |
According to DataSiftConfig.java line 36, i have version 3.0.0 (http.userAgent("DataSift/" + versionPrefix + " Java/v3.0.0");) Also, yes it happends everytime after a couple of hours. Each time we end up after a day or 2 to restart it. I'll try to update the Datasift java api code and see if it solves the issue in the upcomming days ! Thank you ! |
I consistently get the following exception (from an
ErrorListener#exceptionCaught(Throwable)
instance registered withStreamingData
) immediately after subscribing.Workflow is very simple:
Stream
object.And I never seem to get any results, even for interactions containing common terms. Also of possible interest, I get these shortly after:
Using 3.0.0-Beta1. Is this a problem in my environment or a bug with the library.
The text was updated successfully, but these errors were encountered: