Details
-
Type:
Bug
-
Status: Closed (View Workflow)
-
Priority:
Minor
-
Resolution: Cannot Reproduce
-
Component/s: remoting
-
Labels:
-
Environment:* Virtualbox 64-bit Ubuntu
* Java 7, Oracle:
java version "1.7.0_80"
Java(TM) SE Runtime Environment (build 1.7.0_80-b15)
Java HotSpot(TM) 64-Bit Server VM (build 24.80-b11, mixed mode)
* Jenkins installed and run as described in https://wiki.jenkins-ci.org/display/JENKINS/Building+Jenkins, except I checked out the LTS version tag 2.7.4 (remoting v2.60)* Virtualbox 64-bit Ubuntu * Java 7, Oracle: java version "1.7.0_80" Java(TM) SE Runtime Environment (build 1.7.0_80-b15) Java HotSpot(TM) 64-Bit Server VM (build 24.80-b11, mixed mode) * Jenkins installed and run as described in https://wiki.jenkins-ci.org/display/JENKINS/Building+Jenkins, except I checked out the LTS version tag 2.7.4 (remoting v2.60)
-
Similar Issues:
Description
Running a local slave, launched using a command, with the -slaveLog option causes a log file to be written (as expected) with every message printed twice there, like this:
channel startedchannel started channel stoppedchannel stopped
(which is not expected).
The launch script looks as follows (/home/matej/wc/jenkins is path to the Jenkins clone):
#!/bin/bash --norc
export TMP=$(mktemp -d --tmpdir jenkins-XXXXX)
SLAVEJAR_URL=http://localhost:8080/jenkins/jnlpJars/slave.jar
wget -q "${SLAVEJAR_URL:?}" -P "${TMP:?}" 2>&1
_log_dir=/home/matej/wc/jenkins/work/logs
mkdir -p "$_log_dir"
_log=$(mktemp -u "$_log_dir/$(date -u +%Y%m%d-%H%M%S)-XXX")
export JAVA_HOME=/usr/lib/jvm/java-7-oracle
"$JAVA_HOME"/bin/java -jar "$TMP"/slave.jar \
-slaveLog "$_log"
I wondered whether it is just the messages that gets duplicated (perhaps caused by the TeeOutputStream created by the Launcher effectively outputting twice to the identical output stream) or if it is the threads calling Launcher.run() (I cannot imagine how that could happen).
Attachments
Issue Links
- is related to
-
JENKINS-39369 Automatically dump agent logs to the disk
-
- Resolved
-
Activity
slaveLog redirects only the STDERR stream, so it should not happen. In the original request the example shows two different messages. Will check if it happens on my instance
JENKINS-39369 should solve this issue as well by providing a better log management option
I don't see any indication in any recent logs, including many I've been recently examining, that this still occurs. I'm going to go ahead and close this issue. If it still occurs then re-open with additional reproduction information.
Jeff Thompson
added a comment - I don't see any indication in any recent logs, including many I've been recently examining, that this still occurs. I'm going to go ahead and close this issue. If it still occurs then re-open with additional reproduction information. Thanks for getting back to this. I am happy to believe that it was solved by Oleg's changes as commented above. If I happen to observe the erroneous behaviour again with a recent Jenkins version, I will update this ticket as suggested.
Matěj Korvas
added a comment - Thanks for getting back to this. I am happy to believe that it was solved by Oleg's changes as commented above. If I happen to observe the erroneous behaviour again with a recent Jenkins version, I will update this ticket as suggested. 
I have removed the note about potentially related issue of slaves failing to start. It turned out to be solvable by using the -text argument to slave.jar, while messages continue to be printed twice in slave logs.