On 21/04/17 03:18 PM, Andrei Borzenkov wrote:
21.04.2017 22:03, Anton Aylward пишет:
On 21/04/17 12:25 PM, Per Jessen wrote:
L A Walsh wrote:
Per Jessen wrote:
man modules-load.d
not found -- at the place it died, there are no manpages available.
systemd-modules-load.service(8) reads files from the above directories which contain kernel modules to load during boot in a static list. Each configuration file is named in the style of /etc/modules-load.d/program.conf.
The configuration files should simply contain a list of kernel module names to load, separated by newlines. Empty lines and lines whose first non-whitespace character is # or ; are ignored
I see...maybe I'll try that next. Sorry, but that wasn't clear. Thanks for clearing that up.
No problem - when you didn't know about modules-load.d, it's an honest mistake to make.
Let's clarify this: It loads a kernel module. It *ONLY* loads a kernel modules.
In the case of LVM it does not
* load any of the LVM config files * any of the user level LVM libraries or binaries * any of the UDEV rules pertaining to LVM or LVM initialization * any of the scripts that are needed to support any of the above * does not start lvmetad * does not run 'pvscan -a'
It is NOT, repeat NOT a solution to the issue that Linda was raising.
It is exactly the solution to this issue.
I presume that you had not read my drill down where I determined that Linda was correct in that she saw it to be a bug in YaST. it has nothing whatsoever to do with /etc/modules-load.d. Nothing whatsoever. YaST, the isntaller, does know about LVM. If Linda did a plain install with LVM, nothing fancy, no thin pool at install, just as many of us have done, she would have to a basic working system. YaST knows about LVM. The basic install sets up systemd modules that take care of the modules. As I pointed out, dracut build with the LVM dracut module brings in the relevant parts of /sbin and /lib65 and /usr/lib/modules/ for the specific kernel and puts them in the initrd so that the LVM can start up. There is no need for any entry in /etc/load-modules And, as I pointed out in the above bullet list, merely putting an entry in there does not do the other things necessary to have a functioning LVM. However putting the dracut LVM module, which is a completely different thing altogether, q.v. the man pages, does ensure all the other parts are there in the initrd and that they get executed; all the items on my above bullet list are satisfied. Other people have also pointed out that they have a functioning LVM system without an entry in /etc/load-modules. The only way you'll need that is if you did not set up your initrd and systemd to initialise LVM. Even so, the items on my bullet list still need to be satisfied. Some of them will be if you install the relevant RPMs, but not all. Some need to be executed at run-time to set up a *PROPERLY* functioning LVM environment, and merely having an entry in /etc/load-modules does not do that. There is a bug in the installer, in YaST. As I pointed out in separate post, the code for generating a thin pool has a bug in it. This bug is active even in a properly installed LVM, one that boots an otherwise fully functional, robust LVM based system. Having or not having an entry in /etc/load-modules makes no difference. Linda was correct, the bug is in YaST. She just didbn't drill down the way I did to determine what it was, inspect the logs, where it is reported quite celarly. Here, once again, is the bug: **************** LVM_LV_CREATE_FAILED System error code was: -4014 /sbin/lvcreate --zero=y --yes -V 2097512k --name 'vpool1' --thin 'VGMain/LVPool': --zero is supported only with thin pool creation Run 'lvcreate --help' for more information ***************** Please note that this is a YaST bug. When I try creating the 'thin' LVM to use the pool by hand, at the CLI, omitting that "--zero=y", it works just fine. Here's the irony. Its done in the low-level cod in /usr/lib64/libstorage.o If you look in https://github.com/openSUSE/libstorage specifically in https://github.com/openSUSE/libstorage/blob/master/package/libstorage.change... you'll see this bug has been fixed: ======================================================================= Mon Feb 13 11:39:52 CET 2017 - aschnell@suse.com - omit option --zero for lvcreate when creating thin provisioned volumes (bsc#968346) - 2.26.12 ========================================================================= That notice of document for the changes in Tumbleweed for February, if anybody bothered to read it: http://mirror.datto.com/opensuse/tumbleweed/iso/Changes.20170216.txt -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org