#508 Issue closed: fedora21beta diskrestore.sh udevcontrol not found¶
Labels: bug, fixed / solved / done
gdha opened issue at 2014-11-12 15:16:¶
During rear recover we hit an error:
+++ type -p udevadm
+++ udevadm control --stop-exec-queue
+++ udevcontrol stop_exec_queue
/var/lib/rear/layout/diskrestore.sh: line 20: udevcontrol: command not found
For some reason:
layout/prepare/GNU/Linux/10_include_partition_code.sh:type -p udevadm >/dev/null && udevadm control --stop-exec-queue || udevcontrol stop_exec_queue
is not interpreted as we hoped it would
gdha commented at 2014-11-12 15:17:¶
related to issues #480 and #490
tbsky commented at 2014-11-12 17:00:¶
hi:
I just install fedora21 beta server 64bit version. and udevadm
control--stop-exec-queue return 0. so I think maybe the problem only
happens when restore? I will try to figure out what happened.
also I think the error message is ugly and misleading. when I did the
patch, I tried to use minimal code to do it. I will try to improve it..
tbsky commented at 2014-11-12 17:33:¶
hi:
I try to do "rear recover" at another vm and "udevadm control
--stop-exec-queue" works fine and didn't complain. so my first restore
just fine. when I restore second time, I met some problem mentioned by
others (lvm wait for yes, strange I never met this problem with RHEL6),
but udevadm didn't return error code 2, it returns error code 0. maybe
the fedora 21 beta version we used are different? I install fedora21
beta by default (I don't choose anything, let installer decide
everything for me).
tbsky commented at 2014-11-12 17:52:¶
hi:
try to recover several times at original vm, and it seems fine. also I
found something interesting, at first I backup at VM-A and restore to
VM-B, VM-B has a larger hardisk then VM-A. when I do second time restore
to VM-B, I met the "lvm wait yes" problem and grub2 can not install
problem. (first time restore is fine).
if the backup and restore vm disk are the same size, I never got any problem...
gdha commented at 2014-11-18 13:18:¶
@tbsky lvm wait yes" problem is related to issue #493 (scroll down in the issue). Should now be fixed in github... I think we are good to close this on?
tbsky commented at 2014-11-19 05:06:¶
hi gdha:
I just check out github and did find the fix of "lvcreate" command to fix the "lvm wait yes" issue. I haven't try it, but it looks good.
also I find out the fix of "udevadm control --stop-exec-queue". but I don't think it is a proper fix. as I said, the udev problem is not nly related to mdadm only, it is related to many procedure of disk operation. and the new fix only pause udev, but didn't unpause it (I don't know if udevadm trigger/settle will unpause udev, maybe you have try it so these commands will unpause udev?)
the puzzle for me is that I can not reproduce what you saw under fedora 21 beta. in my testing "udevadm control --stop-exec-queue" is working fine under fedora 21 beta. could you tell me what iso/install method to use so I can reproduce what you saw?
gdha commented at 2014-12-05 16:53:¶
@tbsky I prefer to leave it as it as for 1.17 as it is impossible for me to test all the distro's and versions mix. I also moved the milestone to 1.18
tbsky commented at 2014-12-18 15:54:¶
@gdha I create a new pull request, hope it can make the situation better..
gdha commented at 2014-12-24 12:45:¶
@tbsky I can confirm it works well on Fedora21 now. Thanks for the hard work and patience...
gdha commented at 2015-02-16 16:10:¶
added to the release notes so we can close this issue
[Export of Github issue for rear/rear.]