quotes in bash script on windows are being ignored
Hi, I have a jenkins project whose Execute shell command consists of this: set -e # Exit immediately if a simple command exits with a non-zero status set -x verbose export PKG=annotation export SUFFIX=zip ssh -o StrictHostKeyChecking=no -i $SSH_KEY jenk...@docbuilder.mydomain.org rm -f ~/repository/$PKG/*.$SUFFIX If I run this script manually on the windows box, it works fine. That ssh command looks syntactically valid to me (SSH_KEY is defined in the node config) but it fails in jenkins. Since I have verbosity turned on this is the command I see (along with its output) in the jenkins build output: + ssh -o StrictHostKeyChecking=no -i c:/jenkins/docbuilder jenk...@docbuilder.mydomain.org rm -f ~/repository/annotation/*.zip usage: ssh [-1246AaCfgKkMNnqsTtVvXxYy] [-b bind_address] [-c cipher_spec] [-D [bind_address:]port] [-e escape_char] [-F configfile] [-I pkcs11] [-i identity_file] [-L [bind_address:]port:host:hostport] [-l login_name] [-m mac_spec] [-O ctl_cmd] [-o option] [-p port] [-R [bind_address:]port:host:hostport] [-S ctl_path] [-W host:port] [-w local_tun[:remote_tun]] [user@]hostname [command] So it looks like the quotes in my command were somehow stripped. I think those quotes are needed in order for ssh to understand its arguments (otherwise the * is expanded before it gets to ssh, and so on). How can I fix that? I tried escaping the quotes with a backslash and then the command (when echoed) looked like this in the jenkins output: + ssh -o StrictHostKeyChecking=no -i c:/jenkins/docbuilder jenk...@docbuilder.mydomain.org rm -f ~/repository/annotation/*.zip usage: ssh [-1246AaCfgKkMNnqsTtVvXxYy] [-b bind_address] [-c cipher_spec] [-D [bind_address:]port] [-e escape_char] [-F configfile] [-I pkcs11] [-i identity_file] [-L [bind_address:]port:host:hostport] [-l login_name] [-m mac_spec] [-O ctl_cmd] [-o option] [-p port] [-R [bind_address:]port:host:hostport] [-S ctl_path] [-W host:port] [-w local_tun[:remote_tun]] [user@]hostname [command] So the quotes are now present but there is still something wrong with the command as seen by ssh. If I copy and paste that command directly into an sh window on the windows node (a cygwin terminal window where I have exported SSH_KEY), it works. Another thing I should mention is that I'm logged into the windows node as Administrator, but the jenkins slave is running as nt authority\system, a user who does not have the right to log on interactively. Any ideas? Thanks. -- You received this message because you are subscribed to the Google Groups Jenkins Users group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
unable to connect slaves (behind firewalls) to master (in cloud) - this used to work
Hello, Previously I had a Jenkins master running on an EC2 instance in the Amazon cloud, and two slave nodes, one of which was also in EC2 and the other of which is here at my institution, behind a firewall. Both slaves connected to the master using Java Web Start and everything worked fine. Now I am not able to set up new slaves inside my institution--they time out while handshaking with the master. And while my existing node inside the institute continued to work, after I disconnected it and tried to reconnect, it experienced the same problem. I am pretty sure that the problem is being behind the institution firewall, because when I unplugged my laptop from the ethernet cable and connected it to a public wi-fi network, it was able to connect, but when connected by ethernet only it times out during handshaking. The master is running Jenkins ver. 1.532. The slaves are running the slave.jar from http://mymasternode.tld:8080/jnlpJars/slave.jar. To be clear, the slave nodes CAN start a network connection with the master. I've set the master up to use a fixed port (33440) for JNLP slave agents, and made sure that port is made available in my EC2 security group. From a machine at my institution I can do this (slightly edited): $ telnet mymasternode.tld 33440 Trying xx.xx.xx.xx... Connected to mymasternode.tld Escape character is '^]'. So that means in principle I can initiate a socket connection to the master. I try starting a slave agent like this: java -Djava.awt.headless=true -jar slave.jar -jnlpUrl http://mymasternode.tld:8080/computer/laptop/slave-agent.jnlp -secret LONGSECRETCODEGOESHERE I get this: Sep 25, 2013 4:55:55 PM hudson.remoting.jnlp.Main$CuiListener init INFO: Jenkins agent is running in headless mode. Sep 25, 2013 4:55:55 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Locating server among [http://mymasternode.tld:8080/] Sep 25, 2013 4:55:55 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Connecting to mymasternode.tld:33440 Sep 25, 2013 4:55:55 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Handshaking Sep 25, 2013 4:55:56 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Connected On the master in $JENKINS_HOME/logs/slaves/myslavename, I see this in the log: NLP agent connected from /xx.xx.xx.xx ===[JENKINS REMOTING CAPACITY]=== and then after several minutes I see this: ERROR: Connection terminated ha:WC+LCP9b85aBtbiIQSmjNKU4P08vOT+vOD8mVc8DzHWtSE4tKMnMz/PLL0ldFVf2c+b/lb5MDAwVRQxSaBqcITRIIQMEMIIUFgAAckCEiWA=java.net.SocketException: Connection timed out at java.net.SocketInputStream.socketRead0(Native Method) at java.net.SocketInputStream.read(SocketInputStream.java:146) at java.io.BufferedInputStream.fill(BufferedInputStream.java:235) at java.io.BufferedInputStream.read(BufferedInputStream.java:254) at hudson.remoting.FlightRecorderInputStream.read(FlightRecorderInputStream.java:77) at java.io.ObjectInputStream$PeekInputStream.peek(ObjectInputStream.java:2271) at java.io.ObjectInputStream$BlockDataInputStream.peek(ObjectInputStream.java:2564) at java.io.ObjectInputStream$BlockDataInputStream.peekByte(ObjectInputStream.java:2574) at java.io.ObjectInputStream.readObject0(ObjectInputStream.java:1314) at java.io.ObjectInputStream.readObject(ObjectInputStream.java:368) at hudson.remoting.Command.readFrom(Command.java:92) at hudson.remoting.ClassicCommandTransport.read(ClassicCommandTransport.java:71) at hudson.remoting.SynchronousCommandTransport$ReaderThread.run(SynchronousCommandTransport.java:48) Back at the slave, it still says 'connected'. What is going on? OK, so I'm behind a firewall, but I've shown that I can initiate a socket connection, so why do inside-the-firewall slaves behave differently during the handshake than outside-the-firewall slaves which start right up just fine? Maybe something changed on my end but I don't know what it could be so not sure what to ask my IT folks about. Thought maybe someone here could help. Thanks, Dan -- You received this message because you are subscribed to the Google Groups Jenkins Users group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
downstream job set by Parameterized Trigger never runs
Hi, I'm using the Parameterized Trigger plugin. I have a multi-configuration project (project A) which has a post-build action to Trigger parameterized build on other projects, configured to run another multi-configuration project (project B) when the build is stable, and pass B the current build parameters from A. The problem is that the trigger does not seem to work. Project B never runs even though project A builds successfully. In fact even if I change the configuration on A to trigger B when the build is Complete (always trigger), it still does not trigger B. What am I doing wrong? Thanks, Dan -- You received this message because you are subscribed to the Google Groups Jenkins Users group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
Re: downstream job set by Parameterized Trigger never runs
On Tuesday, May 14, 2013 10:09:23 AM UTC-7, Dan Tenenbaum wrote: Hi, I'm using the Parameterized Trigger plugin. I have a multi-configuration project (project A) which has a post-build action to Trigger parameterized build on other projects, configured to run another multi-configuration project (project B) when the build is stable, and pass B the current build parameters from A. The problem is that the trigger does not seem to work. Project B never runs even though project A builds successfully. In fact even if I change the configuration on A to trigger B when the build is Complete (always trigger), it still does not trigger B. What am I doing wrong? If I check the Trigger build without parameters box, project B is indeed run. However, project B (which is currently just a shell script that prints out the current environment with 'env') does not display any of the build parameters from project A. I don't understand what the Trigger build without parameters checkbox is. The whole reason I am running the Parameterized Trigger plugin is to be able to pass A's build parameters to B. If I look at the console output of B, I see originally caused by: Started by upstream project A build number 50 All I really need is that A (in other words, the name of the project that triggered B) but I'm not sure how to get it programmatically. I should mention that A is not a parameterized project in that it does not take any special parameters. The parameters I want to pass to B are just the ordinary ones that Jenkins sets, like JOB_NAME and so forth. OK, it looks like maybe what I want to do is write a properties file in project A and pass that to project B with the Parameters from properties file option. But that doesn't seem to work. I can verify in project A that the properties file is written out and has the correct contents, but in project B I echo out the parameters I'm trying to pass and they are not defined. One clue is that if I check the Don't trigger if any files are missing box, project B does not run. This tells me that Jenkins *thinks* that there is no such properties file in my workspace, but there is. In the top level of the workspace I have a file called job.properties and I've put 'job.properties' (without the quotes) in the Use properties from file box. However, it works if I use Predefined parameters like this: PREV_WORKSPACE=$WORKSPACE So I guess my question is answered. Thanks, Dan Thanks, Dan -- You received this message because you are subscribed to the Google Groups Jenkins Users group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
incorrect parameters passed to downstream build with Parameterized Trigger plugin
Hi, I have two multi-configuration projects, A and B. Each job is configured to run on the same three nodes: the master (linux) and mac and windows slaves. Project A is set up to trigger a build of B when A builds successfully, passing it the following predefined parameters (using the Parameterized Trigger plugin): PREV_WORKSPACE=$WORKSPACE PREV_JOB_NAME=$JOB_NAME PREV_BUILD_TAG=$BUILD_TAG PREV_BUILD_ID=$BUILD_ID PREV_BUILD_DISPLAY_NAME=$BUILD_DISPLAY_NAME For now, project B just calls a shell script that prints out some of these values. On the master, B says that $PREV_WORKSPACE is /var/lib/jenkins/workspace/A which is correct. On the other nodes, though, it ALSO gives the same value for $PREV_WORKSPACE which is WRONG...the workspace path is different on each node. Is this a bug? Can it be fixed? Thanks, Dan -- You received this message because you are subscribed to the Google Groups Jenkins Users group. To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
Re: difficulty connecting to Mac slave
On Wed, May 1, 2013 at 8:12 PM, Kenny Ayers theotherwhitem...@gmail.com wrote: Hey Dan, Oh, I didn't notice you were using javaws. Thank you for the correction. OK, so that ensures that the jenkins web app is always running, but I'm not clear whether that means that the slave will always be running (and in contact with the master). Presumably if the application successfully starts up successfully, that's what you'll get. The Jenkins log file location can vary according to how you started the service. Can you grep out the startup command from the list of running services? There is a logfile option. For example you should see something like: $ ps aux | grep -i jenk java -Dcom.sun.akuma.Daemon=daemonized -Djava.awt.headless=true -DJENKINS_HOME=/var/lib/jenkins -jar /usr/lib/jenkins/jenkins.war --logfile=/var/log/jenkins/jenkins.log --daemon --debug=5 --handlerCountMax=100 --handlerCountMaxIdle=20 I looked over your JNLP startup command, and found that your OSX build agent seems to be trying to connect to localhost. Are you running the master and slave on the same host? If not, is there another instance of Jenkins master running on the build agent that you can kill (presumably this was accidentally setup and left running during testing)? I've highlighted the line below where it shows that the JNLP agent is connecting to localhost: java -jar slave.jar -jnlpUrl http://dhcp151178:8080/computer/pelham/slave-agent.jnlp Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener init INFO: Jenkins agent is running in headless mode. Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Locating server among [http://localhost:8080/, http://dhcp151178:8080/] Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Connecting to localhost:49161 Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Handshaking Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener error SEVERE: The server rejected the connection: Unauthorized access java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) So that seems strange. A few more things to check: is the remote fs root specified correctly, and does it have the correct permissions? Is the Jenkins URL specified correctly on the master node? Maybe getting closer... OK, one thing I guess I did not understand was that the Jenkins webapp does not need to be running on the slave. In fact, shutting it down solved my problem. I was then able to start the slave by doing java -Djava.awt.headless=true -jar slave.jar -jnlpUrl http://master:8080/computer/slave/slave-agent.jnlp -secret secret-key And in fact I could start this command with 'nohup' and pipe its output to a file (and put on the end of it) and I don't need to be logged in on the slave for my builds to work. I'll set up this command to be run automatically at boot time and then everything will 'just work' if the machine is rebooted. Thanks very much for your help. Dan -Kenny On Wednesday, May 1, 2013 6:54:23 PM UTC-7, Dan Tenenbaum wrote: On Wed, May 1, 2013 at 6:48 PM, Kenny Ayers theother...@gmail.com wrote: Hey Dan, I noticed that in part of your reply, you said you had an issue starting the slave, but it looks there is a syntax error here: I also downloaded the JNLP manually and tried this: javaws -Jjava.awt.headless=true slave-agent.jnlp ...but I get: ... Splash: Can't connect to window server.: No such file or directory Wed May 1 15:04:49 pitt.fhcrc.org java[349] Error: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged. Exception in thread main java.lang.InternalError: Can't connect to window server - not enough permissions. Make sure you use -D, not -J. For javaws (as opposed to java) the command to pass an option to the JVM is -J, not -D. Setting up an SSH tunnel from the master to slave should be a workaround for the security error you're getting. Any reason your master has to be outside of your network in the EC2 cloud (I'm guessing this is part of an enterprise support scheme)? I would like all the build machines to be in EC2 but there are no virtual Mac machines because of licensing issues, so in order to build on the Mac I need to contact a machine on our internal network. There are a few ways to run a Jenkins slave on startup for OSX. Here is a link describing how to setup the Jenkins service (not slave): http://jrenard.info/blog/a-quick-but-working-startup-item-for-hudson-for-mac-os-x.html - you should be able to modify the contents of the plist file to run the slave service. OK, so that ensures that the jenkins web app is always running, but I'm not clear whether that means
Re: difficulty connecting to Mac slave
On Tue, Apr 30, 2013 at 6:49 PM, Kenny Ayers theotherwhitem...@gmail.com wrote: You're experiencing at least two issues here. First there's the OSX bug with your window manager permissions. You need to run headless on OSX, as you already determined, so make sure you continue to do that. (The bug is reported here: https://issues.jenkins-ci.org/browse/JENKINS-15697). Add this to the slave command line options: -Djava.awt.headless=true This did the trick. Thanks! Dan Secondarily it seems like perhaps the slave node may not be properly configured within Jenkins? Have you tried a combination of SSH + running headless (perhaps that was your error message you included)? I hope this helps, Kenny On Tuesday, April 30, 2013 5:28:11 PM UTC-7, Dan Tenenbaum wrote: Hi all, I have tried a number of methods to connect to a Mac slave and have had no success. My master is also a mac and it's called dhcp151178. The slave I'm trying to connect to is called pelham. These of course are machines on my internal network. The first method I tried was Launch slave agents on Unix machines via SSH. It looks like the master can connect to the slave via ssh but then I get the following (excerpted; full stack traces at end of message): java.io.IOException: Remote call on pelham failed ... Caused by: java.lang.InternalError: Can't connect to window server - not enough permissions. Then I tried Launch slave agents via Java Web Start And on the slave I navigated to http://dhcp151178:8080/computer/pelham/slave-agent.jnlp This gave me: java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) Also, based on the following post: http://stackoverflow.com/questions/11824486/setting-up-a-hudson-slave-on-windows-xp I tried downloading the jnlp file from the URL above, replacing all instances of 'localhost' with 'dhcp151178' and launching the jnlp with javaws, this either gives me no console output at all (but the slave agent gui says Terminated) or it says: ... Exception in thread main java.lang.InternalError: Can't connect to window server - not enough permissions. I tried starting in headless mode, both as jenkins and as my regular admin user. I did this on the slave: java -jar slave.jar -jnlpUrl http://dhcp151178:8080/computer/pelham/slave-agent.jnlp Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener init INFO: Jenkins agent is running in headless mode. Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Locating server among [http://localhost:8080/, http://dhcp151178:8080/] Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Connecting to localhost:49161 Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Handshaking Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener error SEVERE: The server rejected the connection: Unauthorized access java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) There is no security set up on the master, so I'm not sure what the unauthorized access is about, and there is no -auth username:password that I can add to that command line. I'm not sure how to set up Launch slave via execution of command on the Master because I can't find much documentation for that. So I'm stuck. I was able to set up a windows slave so I know that it is possible. But the Mac slave does not want to start. Can you help? Thanks, Dan Addenda: full stack trace from trying to launch slave agents on the master using ssh: 04/30/13 17:23:29] [SSH] Opening SSH connection to pelham:22. [04/30/13 17:23:29] [SSH] Authentication successful. [04/30/13 17:23:29] [SSH] The remote users environment is: BASH=/bin/bash BASH_ARGC=() BASH_ARGV=() BASH_EXECUTION_STRING=set BASH_LINENO=() BASH_SOURCE=() BASH_VERSINFO=([0]=3 [1]=2 [2]=48 [3]=1 [4]=release [5]=x86_64-apple-darwin10.0) BASH_VERSION='3.2.48(1)-release' DIRSTACK=() EUID=216 GROUPS=() HOME=/Users/Shared/Jenkins HOSTNAME=pelham.fhcrc.org HOSTTYPE=x86_64 IFS=$' \t\n' LOGNAME=jenkins MACHTYPE=x86_64-apple-darwin10.0 MAIL=/var/mail/jenkins OPTERR=1 OPTIND=1 OSTYPE=darwin10.0 PATH=/usr/bin:/bin:/usr/sbin:/sbin PPID=545 PS4='+ ' PWD=/Users/Shared/Jenkins SHELL=/bin/bash SHELLOPTS=braceexpand:hashall:interactive-comments SHLVL=1 SSH_CLIENT='140.107.151.178 56267 22' SSH_CONNECTION='140.107.151.178 56267 140.107.150.101 22' TERM=dumb UID=216 USER=jenkins _=bash [04/30/13 17:23:29] [SSH] Checking java version of java [04/30/13 17:23:29] [SSH] java -version returned 1.6.0_45. [04/30/13 17:23:29] [SSH] Starting sftp client. [04/30/13 17:23:29] [SSH] Copying latest slave.jar... [04/30/13 17:23:29
Re: difficulty connecting to Mac slave
On Wed, May 1, 2013 at 10:04 AM, Dan Tenenbaum danda...@dandante.com wrote: On Tue, Apr 30, 2013 at 6:49 PM, Kenny Ayers theotherwhitem...@gmail.com wrote: You're experiencing at least two issues here. First there's the OSX bug with your window manager permissions. You need to run headless on OSX, as you already determined, so make sure you continue to do that. (The bug is reported here: https://issues.jenkins-ci.org/browse/JENKINS-15697). Add this to the slave command line options: -Djava.awt.headless=true This did the trick. Thanks! Dan Secondarily it seems like perhaps the slave node may not be properly configured within Jenkins? Have you tried a combination of SSH + running headless (perhaps that was your error message you included)? I still have a problem, because I ultimately I want to have a scenario where the master is in the cloud (EC2) and the Mac slave is on my internal network, behind a firewall. So I won't be able to use ssh in this scenario to connect from the master to the slave. It sounds like I will have to start the slave agent on the slave. But that isn't working, even with the headless option to the JVM, I am getting $ java -Djava.awt.headless=true -jar slave.jar -jnlpUrl http://dhcp151178:8080/computer/pitt/slave-agent.jnlp -secret secret hidden May 1, 2013 2:59:10 PM hudson.remoting.jnlp.Main$CuiListener error SEVERE: The server rejected the connection: Unauthorized access java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) This happens even if I provide the -secret suggested to me on the slave's page on the master's Jenkins. I get the same error if I try and start the JNLP from a browser, after logging in to jenkins on the master as a user with admin rights. I also downloaded the JNLP manually and tried this: javaws -Jjava.awt.headless=true slave-agent.jnlp ...but I get: ... Splash: Can't connect to window server.: No such file or directory Wed May 1 15:04:49 pitt.fhcrc.org java[349] Error: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged. Exception in thread main java.lang.InternalError: Can't connect to window server - not enough permissions. ... I thought the java.awt.headless switch was supposed to take care of this. I have added that switch to the master's config of this node. I *do* now have security on the master, so is there a way to pass a username/password when I try and invoke the JNLP from the command line? I assume that's what the -secret option is supposed to do. I am able to start the slave from the master using both Launch slave agents on Unix machines via SSH and Launch slave via execution of command on the Master (using the command ssh jenkins@nameofslave java -Djava.awt.headless=true -jar /Users/Shared/Jenkins/Home/war/WEB-INF/slave.jar) BUT, I cannot establish communication between the slave and master if the master is outside my network firewall. There is an ssh gateway and I'll investigate whether tunneling is a possibility. But I'd like to be able to start the slave agent--that's my real problem. Also I'm wondering if there is a way to start the slave agent on the slave without being logged in. Because what if the slave machine reboots, there is no guarantee that someone will log into it and start the appropriate Java Web Start. There is probably a way to automate that but I'd prefer if the necessary processes ran as daemons so being logged in was not necessary. Thanks in advance, Dan I hope this helps, Kenny On Tuesday, April 30, 2013 5:28:11 PM UTC-7, Dan Tenenbaum wrote: Hi all, I have tried a number of methods to connect to a Mac slave and have had no success. My master is also a mac and it's called dhcp151178. The slave I'm trying to connect to is called pelham. These of course are machines on my internal network. The first method I tried was Launch slave agents on Unix machines via SSH. It looks like the master can connect to the slave via ssh but then I get the following (excerpted; full stack traces at end of message): java.io.IOException: Remote call on pelham failed ... Caused by: java.lang.InternalError: Can't connect to window server - not enough permissions. Then I tried Launch slave agents via Java Web Start And on the slave I navigated to http://dhcp151178:8080/computer/pelham/slave-agent.jnlp This gave me: java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) Also, based on the following post: http://stackoverflow.com/questions/11824486/setting-up-a-hudson-slave-on-windows-xp I tried downloading the jnlp file from the URL above, replacing all instances of 'localhost' with 'dhcp151178' and launching the jnlp with javaws, this either gives me
Re: difficulty connecting to Mac slave
On Wed, May 1, 2013 at 6:48 PM, Kenny Ayers theotherwhitem...@gmail.com wrote: Hey Dan, I noticed that in part of your reply, you said you had an issue starting the slave, but it looks there is a syntax error here: I also downloaded the JNLP manually and tried this: javaws -Jjava.awt.headless=true slave-agent.jnlp ...but I get: ... Splash: Can't connect to window server.: No such file or directory Wed May 1 15:04:49 pitt.fhcrc.org java[349] Error: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged. Exception in thread main java.lang.InternalError: Can't connect to window server - not enough permissions. Make sure you use -D, not -J. For javaws (as opposed to java) the command to pass an option to the JVM is -J, not -D. Setting up an SSH tunnel from the master to slave should be a workaround for the security error you're getting. Any reason your master has to be outside of your network in the EC2 cloud (I'm guessing this is part of an enterprise support scheme)? I would like all the build machines to be in EC2 but there are no virtual Mac machines because of licensing issues, so in order to build on the Mac I need to contact a machine on our internal network. There are a few ways to run a Jenkins slave on startup for OSX. Here is a link describing how to setup the Jenkins service (not slave): http://jrenard.info/blog/a-quick-but-working-startup-item-for-hudson-for-mac-os-x.html - you should be able to modify the contents of the plist file to run the slave service. OK, so that ensures that the jenkins web app is always running, but I'm not clear whether that means that the slave will always be running (and in contact with the master). Does the master node have any information in its log when OSX has the security crash? Any idea where I should look? I'm not sure where jenkins' logs are. The master is running on OSX as well if that helps you answer the question. Have you considered setting up a test instance of Jenkins and debugging in real-time? I realize this is a bit more technical than most users want to get, but if you've got some understanding of Java it is fairly simple to do using NetBeans: https://groups.google.com/forum/?fromgroups=#!topic/jenkinsci-dev/9a4F7jN1JNI I'll try it if other avenues don't work out. Thanks, Dan Hope this helps some, Kenny On Wednesday, May 1, 2013 4:47:46 PM UTC-7, Dan Tenenbaum wrote: On Wed, May 1, 2013 at 10:04 AM, Dan Tenenbaum dand...@dandante.com wrote: On Tue, Apr 30, 2013 at 6:49 PM, Kenny Ayers theother...@gmail.com wrote: You're experiencing at least two issues here. First there's the OSX bug with your window manager permissions. You need to run headless on OSX, as you already determined, so make sure you continue to do that. (The bug is reported here: https://issues.jenkins-ci.org/browse/JENKINS-15697). Add this to the slave command line options: -Djava.awt.headless=true This did the trick. Thanks! Dan Secondarily it seems like perhaps the slave node may not be properly configured within Jenkins? Have you tried a combination of SSH + running headless (perhaps that was your error message you included)? I still have a problem, because I ultimately I want to have a scenario where the master is in the cloud (EC2) and the Mac slave is on my internal network, behind a firewall. So I won't be able to use ssh in this scenario to connect from the master to the slave. It sounds like I will have to start the slave agent on the slave. But that isn't working, even with the headless option to the JVM, I am getting $ java -Djava.awt.headless=true -jar slave.jar -jnlpUrl http://dhcp151178:8080/computer/pitt/slave-agent.jnlp -secret secret hidden May 1, 2013 2:59:10 PM hudson.remoting.jnlp.Main$CuiListener error SEVERE: The server rejected the connection: Unauthorized access java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) This happens even if I provide the -secret suggested to me on the slave's page on the master's Jenkins. I get the same error if I try and start the JNLP from a browser, after logging in to jenkins on the master as a user with admin rights. I also downloaded the JNLP manually and tried this: javaws -Jjava.awt.headless=true slave-agent.jnlp ...but I get: ... Splash: Can't connect to window server.: No such file or directory Wed May 1 15:04:49 pitt.fhcrc.org java[349] Error: kCGErrorFailure: Set a breakpoint @ CGErrorBreakpoint() to catch errors as they are logged. Exception in thread main java.lang.InternalError: Can't connect to window server - not enough permissions. ... I thought the java.awt.headless switch was supposed to take care of this. I have added that switch to the master's config of this node. I
Re: difficulty connecting to Mac slave
On Tue, Apr 30, 2013 at 5:55 PM, Scott Evans milwrd...@gmail.com wrote: Dan, I'm not a mac person so can't help you in detail, but can you, from your mac, browse to your Jenkins server with a web browser? Yes. It almost sounds like your slave mac isn't giving the jenkins slave process enough permissions to reach out, or there is some sort of firewall blocking connections to the jenkins server on port 8080 perhaps. There is no firewall. To me it seems like the slave can contact the master, but is having an authorization problem once it does. Dan Scott On Tue, Apr 30, 2013 at 7:28 PM, Dan Tenenbaum danda...@gmail.com wrote: Hi all, I have tried a number of methods to connect to a Mac slave and have had no success. My master is also a mac and it's called dhcp151178. The slave I'm trying to connect to is called pelham. These of course are machines on my internal network. The first method I tried was Launch slave agents on Unix machines via SSH. It looks like the master can connect to the slave via ssh but then I get the following (excerpted; full stack traces at end of message): java.io.IOException: Remote call on pelham failed ... Caused by: java.lang.InternalError: Can't connect to window server - not enough permissions. Then I tried Launch slave agents via Java Web Start And on the slave I navigated to http://dhcp151178:8080/computer/pelham/slave-agent.jnlp This gave me: java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) Also, based on the following post: http://stackoverflow.com/questions/11824486/setting-up-a-hudson-slave-on-windows-xp I tried downloading the jnlp file from the URL above, replacing all instances of 'localhost' with 'dhcp151178' and launching the jnlp with javaws, this either gives me no console output at all (but the slave agent gui says Terminated) or it says: ... Exception in thread main java.lang.InternalError: Can't connect to window server - not enough permissions. I tried starting in headless mode, both as jenkins and as my regular admin user. I did this on the slave: java -jar slave.jar -jnlpUrl http://dhcp151178:8080/computer/pelham/slave-agent.jnlp Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener init INFO: Jenkins agent is running in headless mode. Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Locating server among [http://localhost:8080/, http://dhcp151178:8080/] Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Connecting to localhost:49161 Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener status INFO: Handshaking Apr 30, 2013 5:19:02 PM hudson.remoting.jnlp.Main$CuiListener error SEVERE: The server rejected the connection: Unauthorized access java.lang.Exception: The server rejected the connection: Unauthorized access at hudson.remoting.Engine.onConnectionRejected(Engine.java:257) at hudson.remoting.Engine.run(Engine.java:232) There is no security set up on the master, so I'm not sure what the unauthorized access is about, and there is no -auth username:password that I can add to that command line. I'm not sure how to set up Launch slave via execution of command on the Master because I can't find much documentation for that. So I'm stuck. I was able to set up a windows slave so I know that it is possible. But the Mac slave does not want to start. Can you help? Thanks, Dan Addenda: full stack trace from trying to launch slave agents on the master using ssh: 04/30/13 17:23:29] [SSH] Opening SSH connection to pelham:22. [04/30/13 17:23:29] [SSH] Authentication successful. [04/30/13 17:23:29] [SSH] The remote users environment is: BASH=/bin/bash BASH_ARGC=() BASH_ARGV=() BASH_EXECUTION_STRING=set BASH_LINENO=() BASH_SOURCE=() BASH_VERSINFO=([0]=3 [1]=2 [2]=48 [3]=1 [4]=release [5]=x86_64-apple-darwin10.0) BASH_VERSION='3.2.48(1)-release' DIRSTACK=() EUID=216 GROUPS=() HOME=/Users/Shared/Jenkins HOSTNAME=pelham.fhcrc.org HOSTTYPE=x86_64 IFS=$' \t\n' LOGNAME=jenkins MACHTYPE=x86_64-apple-darwin10.0 MAIL=/var/mail/jenkins OPTERR=1 OPTIND=1 OSTYPE=darwin10.0 PATH=/usr/bin:/bin:/usr/sbin:/sbin PPID=545 PS4='+ ' PWD=/Users/Shared/Jenkins SHELL=/bin/bash SHELLOPTS=braceexpand:hashall:interactive-comments SHLVL=1 SSH_CLIENT='140.107.151.178 56267 22' SSH_CONNECTION='140.107.151.178 56267 140.107.150.101 22' TERM=dumb UID=216 USER=jenkins _=bash [04/30/13 17:23:29] [SSH] Checking java version of java [04/30/13 17:23:29] [SSH] java -version returned 1.6.0_45. [04/30/13 17:23:29] [SSH] Starting sftp client. [04/30/13 17:23:29] [SSH] Copying latest slave.jar... [04/30/13 17:23:29] [SSH] Copied 284,251 bytes. Expanded the channel window size to 4MB [04/30/13 17:23:29] [SSH] Starting slave process: cd '/Users
