Read timed out

Click on the to mark the solution that helps you, Samebug will learn from it.
As a community member, you’ll be rewarded for you help.
  1. 0

    Unable to get the parse value out of multi-line logs in logstash

    Stack Overflow | 11 months ago | omer khalid Read timed out
  2. 0

    VCS patch building intermittently failing on clearcase. – TeamCity | 10 months ago Read timed out
  3. 0

    Axis2: Timeout when using setUseSeparateListener(true)

    Coderanch | 7 years ago | emengy bh Read timed out
  4. Speed up your debug routine!

    Automated exception search integrated into your IDE

  5. 0

    0001597: Overlapping database pull processes after connection outage - Default pull.thread.per.server.count=1 - SymmetricDS Issues | 1 year ago
    org.jumpmind.exception.IoException: Read timed out
  6. 0

    After upgrading recently to a version of Hudson that includes the Hudson CLI, we started seeing catastrophic heap exhaustion when the "groovysh" command was used. Our Hudson servers are configured to use only the HTTP-based communication method for the CLI, not the ad hoc TCP port method. Digging around a bit, it turns out to be related to read timeouts on the Channel reader. In our case, the HTTP connection timeout was set to 20 seconds, so after that time we would see the CLI get disconnected from Hudson. The server logs show an exception like this: 2009-10-02 12:50:59,786 ERROR [hudson.remoting.Channel] I/O error in channel HTTP full-duplex channel 552692ad-f7ab-4ed2-8eb6-5be614fdcefb Read timed out at Method) at at org.apache.coyote.http11.InternalInputBuffer.fill( at org.apache.coyote.http11.InternalInputBuffer$InputStreamInputBuffer.doRead( at org.apache.coyote.http11.filters.ChunkedInputFilter.readBytes( at org.apache.coyote.http11.filters.ChunkedInputFilter.parseChunkHeader( at org.apache.coyote.http11.filters.ChunkedInputFilter.doRead( at org.apache.coyote.http11.InternalInputBuffer.doRead( at org.apache.coyote.Request.doRead( at org.apache.catalina.connector.InputBuffer.realReadBytes( at org.apache.tomcat.util.buf.ByteChunk.substract( at org.apache.catalina.connector.InputBuffer.readByte( at at$PeekInputStream.peek( at$BlockDataInputStream.peek( at$BlockDataInputStream.peekByte( at at at hudson.remoting.Channel$ This happens in the Channel reader thread, but in the pool thread that is serving the HTTP request, the Groovysh repl loop continues, continuing to look for new commands from the user. Each time it does this, it ends up getting down to Channel.send where an exception is thrown to indicate that the channel is closed. In the case of writes/flush, this exception is swallowed by the PrintStream and PrintWriter wrappers at various levels of the output chain (because they swallow ALL IOExceptions, per their contract). In the case of reads, the IOException ends up being throw up to Groovy's where it is caught as a Throwable and passed to the configured errorHook. The problem is that the errorHook then tries to write the error message out to stderr. This fails, but because of the behavior of PrintStream and PrintWriter, the IOException from Channel.send doesn't get thrown. This allows the loop in to continue, calling readLine again to repeat the cycle. While this is happening, each attempt to read from the ProxyInputStream creates a Chunk request that gets put into Channel.pendingCalls. Since these requests will never be serviced they just sit there, chewing up the heap until it is all gone. I have a patch that fixes this, inasmuch as it can be fixed. Instead of relying on the default errorHook, it uses a custom errorHook that watches for ChannelClosedException. If it sees it, it rethrows it, otherwise it calls the default error hook. This causes the exception to propagate out of and eventually causes the GroovyshCommand to terminate with a failure. Once I fixed that, however, I noticed that FullDuplexHttpChannel and ProxyInputStream$Chunk instances were still being left on the heap. I tracked that down to a shutdown hook being added in Groovy's RecordCommand. Since that hook object has a reference to the Groovysh instance, it transitively holds onto the Channel via, IO.out, and IO.err. My fix for this second problem may be more controversial; I decided to remove the "load", "save", and "record" commands. Besides the shutdown hook, the reason for this is that these three commands are meant to interact with local files, but in the case of the Hudson/Groovy CLI these files will actually be looked for on the server, not the client. I could see having usable variants of "load", "save", and "record" in the Hudson/Groovy CLI, but I don't think the stock ones fit the bill. And last but not least, my patch adds a predefined binding for the "hudson" variable to hudson.model.Hudson.getInstance() and adds a predefined import for "hudson.model.*". I know this potentially three separate issues, so please let me know if I need to split up the patch.

    Jenkins JIRA | 8 years ago | Mike Dillon Read timed out

  1. rp 1 times, last 4 months ago
  2. iridic 1 times, last 5 months ago
  3. Bardh 14 times, last 11 months ago
Not finding the right solution?
Take a tour to get the most out of Samebug.

Tired of useless tips?

Automated exception search integrated into your IDE

Root Cause Analysis


    Read timed out

  2. Java RT
    1. Method)
    1 frame