Hide Forgot
Description of problem: gfs_fsck is missing from release package. When you run gfs2_convert,Step 2 tells you to run gfs_fsck. gfs_fsck - file not found.... If you skip this step gfs2_convert fails to convert filesystem to GFS2 Version-Release number of selected component (if applicable): 0.1.62 How reproducible: In RHEL6 install cluster with all components. add drives with GFS1 file system. Try to run gfs_fsck. It fails Steps to Reproduce: 1.Install RHEL6 with all cluster components 2. gfs_fsck 3. not found .... Actual results: gfs_fsck not found. Expected results: gfs_fsck checks volume and correct it. Additional info: Happy April 1st! But I hope it is easy to fix and include in updated package.
In fact, the entire gfs-utils package is not shipped with RHEL6. This was done by design because GFS1 is deprecated for the release. I think the user documentation states that gfs_fsck should be run before the system is upgraded to RHEL6 or something to that effect. For situations like this, we already have bug #675723 which, when done, will allow you to run fsck.gfs2 on your gfs1 file system. This is preferred anyway, since fsck.gfs2 is faster, more robust and more accurate than gfs_fsck. Unfortunately, the code is not yet finished. My prototype is nearly complete but still needs some work, and after that, a complete testing cycle. (If you want, I can provide a preliminary version that you can help me test), but it won't be available for a couple weeks. With that fix, we will change the notice on the gfs2_convert tool to run fsck.gfs2, not gfs_fsck, before converting.
Robert, Sure I would be happy to help you and test this prototype. Could you let me know when you'll have it ready? Thank you Oleg
Hi Oleg, Here is a prototype you can try: http://people.redhat.com/rpeterso/Experimental/RHEL6.x/fsck.gfs2 Please remember that this prototype has not undergone the Red Hat quality assurance process, so it should be treated as experimental. We are not responsible for problems that may result from using this code. It still needs a lot more testing, and that is planned. I have, however, run it on a customer's metadata set and it seemed to work properly. Let me know how it goes.