Created attachment 1510818 [details] config file +++ This bug was initially created as a clone of Bug #1655443 +++ Description of problem: --------------------------- Gdeploy execution did not stop when there was an error encountered with VDO volume creation.The vdo volume creation failed in one of the host still gdeploy went ahead and started created physical volume. Version-Release number of selected component: ----------------------------------------------- rhvh-4.2.8.0-0.20181127 glusterfs-cli-3.12.2-29.el7rhgs.x86_64 How reproducible: ------------------ Everytime Steps to Reproduce: ------------------- 1.Log in to the cockpit UI and start with gluster deployment. Actual results: ------------------- The gdeploy execution proceeded when error was encountered. Expected results: ------------------ The gdeploy execution should halt . Additional Info: ------------------ Created a config file to create a vdo volume and then if that is successful it should create a test file.The vdo creation failed still the test file was generated.
Sachi, is there a plan to address this bug? If not, can you close with comments
(In reply to Sahina Bose from comment #2) > Sachi, is there a plan to address this bug? If not, can you close with > comments I'm sorry, this missed somehow in my gdeploy query. I should address this, will do it for the upcoming release. It shouldn't continue in case of errors.
Mugdha, actually it is not skipping error but it is not running the vdo creation at all. That is because slabsize is not mentioned in the config file, because of which it is not populating the vdo variables into inventory. We did not hit this in our rhhi installation because we always provide slabsize in config. gdeploy should add default values for slabsize. This is a good find, thank you. The patch: https://github.com/gluster/gdeploy/commit/49f4e1efadaf31 fixes the issue.
Some of the configurations: [hosts] 10.70.43.52 [vdo] action=create devices=/dev/sde,/dev/sdd names=test_vol_1,test_vol_2 logicalsize=50G,100G slabsize=1G ignore_vdo_errors=no [shell] action=execute command=echo 'pass' > /root/test ===================== [hosts] 10.70.43.52 [vdo] action=create devices=/dev/sde,/dev/sdd names=test_vol_1,test_vol_2 ignore_vdo_errors=no [shell] action=execute command=echo 'pass' > /root/test
Sac, I see the fix is available for gdeploy, can we also have this fix in gluster-ansible roles ?
sas, this fix is available in gluster-ansible we assume default slabsize in gluster-ansible. Ref: https://github.com/gluster/gluster-ansible-infra/blob/master/roles/backend_setup/tasks/vdo_create.yml#L15
(In reply to Sachidananda Urs from comment #10) > sas, this fix is available in gluster-ansible we assume default slabsize in > gluster-ansible. > Ref: > https://github.com/gluster/gluster-ansible-infra/blob/master/roles/ > backend_setup/tasks/vdo_create.yml#L15 If the fix for this issue is already available in gluster-ansible, then we can have this bug targeted for RHGS 3.4.4 Is that ok ?
(In reply to SATHEESARAN from comment #11) > (In reply to Sachidananda Urs from comment #10) > > sas, this fix is available in gluster-ansible we assume default slabsize in > > gluster-ansible. > > Ref: > > https://github.com/gluster/gluster-ansible-infra/blob/master/roles/ > > backend_setup/tasks/vdo_create.yml#L15 > > If the fix for this issue is already available in gluster-ansible, > then we can have this bug targeted for RHGS 3.4.4 > > Is that ok ? This is gdeploy bug, but component put wrongly.
Closing this bug, as this fix will not be consumed by cockpit. Cockpit will be using gluster-ansible roles for the upcoming RHV releases.