Red Hat Bugzilla – Bug 1021152
3.3.1 beta, add vdsm host fails with error: Unexpected connection termination
Last modified: 2013-10-20 17:16:54 EDT
Created attachment 814092 [details] portion of engine.log from centos 6.4 / ovirt 3.3.1 beta engine Description of problem: Tested with F19 engine and with CentOS 6.4 engine, running 3.3.1 beta packages. On adding a new host, add operation fails due to "java.io.IOException: Unexpected connection termination." Fuller log passage attached. I confirmed that both machines have resolvable DNS names, both machines can see each other, ssh to each other, selinux permissive, tried both w/ firewalls running & not running. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
stderr: bash: tar: command not found *** This bug has been marked as a duplicate of bug 986882 ***
Created attachment 814309 [details] engine log excerpt from attempt to add f19 host (with tar installed) to f19 engine
(In reply to Jason Brooks from comment #2) > Created attachment 814309 [details] > engine log excerpt from attempt to add f19 host (with tar installed) to f19 > engine Please attach full log. Have you installed tar on host or on engine?
Created attachment 814310 [details] log from f19 engine tar is installed on the host and on the engine -- I installed ovirt-release-fedora on both machines, and tar is a dep of that pkg
(In reply to Jason Brooks from comment #4) > Created attachment 814310 [details] > log from f19 engine > > tar is installed on the host and on the engine -- I installed > ovirt-release-fedora on both machines, and tar is a dep of that pkg This is different issue. Please file separate bug, or better discuss first at user mailing list... What do you have at /var/log/messages at host side? Do you have valid tar at /var/cache/ovirt-engine/ovirt-host-deploy.tar, should be non empty with lots of files. What do you get while executing the following command at engine: ( cat /var/cache/ovirt-engine/ovirt-host-deploy.tar && cat ) | ssh root@10.10.171.201 'umask 0077; MYTMP="$(mktemp -t ovirt-XXXXXXXXXX)"; trap "chmod -R u+rwX \"${MYTMP}\" > /dev/null 2>&1; rm -fr \"${MYTMP}\" > /dev/null 2>&1" 0; rm -fr "${MYTMP}" && mkdir "${MYTMP}" && tar --warning=no-timestamp -C "${MYTMP}" -x && "${MYTMP}"/setup DIALOG/dialect=str:machine DIALOG/customization=bool:True'
All right, this report wasn't really about tar -- after having this problem on F19, I tried on CentOS, and forgot to install the ovirt-release pkg. That pkg pulls in tar, so that's not really a problem. Host side, /var/log/messages has NetworkManager activity -- NetworkManager appears to request a new dhcp address when the add host operation begins. I thought maybe could have been disrupting the communications, so I stopped the NetworkManager service, tried add again, no more NetworkManager activity, but the same error occurred engine-side, and on the host side, in /var/log/messages: Oct 20 16:27:22 dhcp-10-171-201 systemd-logind[450]: New session 55 of user root. Oct 20 16:27:23 dhcp-10-171-201 systemd-logind[450]: Removed session 55. Oct 20 16:27:23 dhcp-10-171-201 systemd-logind[450]: New session 56 of user root. Oct 20 16:27:23 dhcp-10-171-201 systemd-logind[450]: Removed session 56. Engine side, ovirt-host-deploy.tar is full of files. Running that command on the engine prompts me for the root pw of the host, then asks me to answer yes/no, and then fails with "ERROR Failed to execute stage 'Initializing': need more than 1 value to unpack"
please move discussion to users@ovirt.org, bugzilla is not the place for that, especially this bug which is duplicate per your original report. please also be consistent in the environment you are using, from this thread I do not understand what OS you are using for engine and what OS you are using for host.