Description of problem: ---------------------- Failures creating an LVM logical volume of type GFS2 using luci. Version-Release number of selected component: -------------------------------------------- luci-0.12.0.7.el5 How reproducible: ---------------- Consistently Steps to Reproduce (in luci): ---------------------------- 1. on cmd line, pvcreate to init 3 /dev/mapper devs for LVM use 2. in luci, create volume group using the 3 devs 3. create logical volume using volgrp created in step 2 Actual results: -------------- luci appears to successfully create the volume group using the 3 devs but then trying to create a logical volume w/GFS2 fails. Each attempt just keeps spinning in luci and eventually has to be stopped. luci has all the required info to move forward with the attempt (mount points, unique GFS name, etc.) http://irish.lab.bos.redhat.com/pub/tim/projects/webserver/screenshots/luci-logvol5.png On the server, the first attempt hung the rhel5.2 OS and I had to bounce it. After rebooting, there was no logical volume created but the vol group was still present so I tried a few more times. luci never returns which I eventually stop after 20 min. The last attempt looks like the log vol was created this time (at least luci finds it and says it's GFS2 with no mounting info) but no entry has been made in /etc/fstab and it does not look like GFS of any type has been applied to the cluster (the cluster.conf has not been updated and contains no GFS references). http://irish.lab.bos.redhat.com/pub/tim/projects/webserver/screenshots/luci-logvol7.png I try to modify this existing logical volume to give it the mount info ... http://irish.lab.bos.redhat.com/pub/tim/projects/webserver/screenshots/luci-logvol9.png ... but this ends in the same lack of return from luci and the cluster.conf file is never changed. I have seen the following error a couple times (but not all) attempting to create the GFS2 file system ... An error has occured while committing changes: <?xml version="1.0" ?> <dict name="result"> <var name="result" type="num" value="True"/> <result name="success" value="true"/> </dict> Expected results: ---------------- - luci creates logical volume group (inside of 10 minute luci inactivity timeout) - volume successfully mounted as GFS2 - volume mount info in /etc/fstab - other cluster nodes should have same mount and fstab entries Additional info: --------------- Same results with and without SELinux enforced.
Additional info: --------------- further testing shows creation of any type of logical volume (GFS*, Extended FS, etc.) fails. If GFS* is chosen, I get the error(s) listed above. Extended FS option does nothing when Create button is clicked. Reset button and pulldown menu works but Create does not prompt for verification or otherwise move forward. I did successfully test this with a single device before trying multiple. Will try that with one of the 3 devs I'm using now in the volgrp.
Single device volgrp still fails to commit the changes. This error is spit out twice for each attempt to create a logvol. An error has occured while committing changes: <?xml version="1.0" ?> <dict name="result"> <var name="result" type="num" value="True"/> <result name="success" value="true"/> </dict>
This time a new error was thrown trying to create a single disk volume group. I was directed to a luci page that read ... Site error This site encountered an error trying to fulfill your request. The errors were: Error Type TypeError Error Value unsubscriptable object Request made at 2008/11/03 15:27:15.923 US/Eastern ... and the volume group was not created. A second attempt generate a pop-up ... An error has occured while committing changes: Reason for failure (as reported by et-virt08-ic.lab.bos.redhat.com): invalid number of mapper_templates returned ... but this time the volume group was successfully created. If anyone opens this code, it would be good if we spelled the word "occurred" correctly in our conga pop-ups.
Attempting to create a single disk volume group, this pop-up is seen prior to the luci web page described in Comment #3 above. We've actually coded in a pop-up telling the user to file a bug against luci? http://irish.lab.bos.redhat.com/pub/tim/projects/webserver/screenshots/Vol2/luciBug1.png
Previous attempts have failed but one (even though it still generated an error) did create the logical volume, it just failed to mount it or enter it into the fstab files as instructed. When I tried to modify the logical volume to add the mount point info by hand, it complained about /bin/mount (haven't seen that one before). http://irish.lab.bos.redhat.com/pub/tim/projects/webserver/screenshots/Vol2/luciBug2.png
*** Bug 472938 has been marked as a duplicate of this bug. ***
I've a similar problem with Centos 5.5 (luci-0.12.2) while creating a storage cluster with 2 nodes and using clvmd. In my workflow I've created a LVM-Partition on both nodes and initiated as physical volume (pvcreate). In luci a can select this partition to create a VolumeGroup. In the webpage appears 'committing changes' until the site timeouts. After that in luci probing storage fails without any error message; the page appears again and again without result. The volume group was created with clusterd attribute on the first node. In /var/lib/ricci/queue remains a batch file with xml-code of the creating job. Restarting of ricci (both nodes) and luci didn't help. I suppose there is a problem between luci and ricci although firewall and selinux are disabled:
Hi Heinrich! We appreciate the input on this bug and additional data, however we would encourage you to contact the CentOS support teams (forums, etc) for assistance with CentOS packages. Bugzilla is not a support tool, even for Red Hat packages. Issues affecting Red Hat software should be directed to the Red Hat support team (www.redhat.com/support). Thanks for your understanding, Jeremy West Red Hat Support Supervisor
Development Management has reviewed and declined this request. You may appeal this decision by reopening this request.