Bug 1341484

*** Bug 1346253 has been marked as a duplicate of this bug. ***

gdeploy should provide the option to abort the installation when the particular section fails.

What happens with HCI installation, if something fails, gdeploy doesn't exit there, but continues to the next step, which lead to total misconfiguration.

Take for example, if the creation of vg fails, then there is no point in creation of lv1 ( thick-pool ), lv2 ( thin-pool ), lv3 ( thin lv), lv4 (thin-lv )

Expected - 
Provide an option in each of the section - 'abort_on_failure=<val>' - where val can be 0 or 1.

During workshop, we found the lack of this feature a big issue. If at any step, actions fail - for instance LV creation fails - gdeploy continues ahead but bricks are mounted at root rather than created LV.

We need this RFE to be taken in.

*** Bug 1346652 has been marked as a duplicate of this bug. ***

Hi Sac,

   what is the fix here? Do i need to add some variable @ all sections so that it does not continue to other steps when the previous one fails ? Could you please add the patch to this bug which fixes this issue?

Thanks
kasturi

Hi Kasturi,

I've documented the flags under every feature and can be found at: http://gdeploy.readthedocs.io/en/latest/features.html

This is how it works. Under lv section you can set

[lv]
...
...
...
ignore_lv_errors=no

[yum]
.
.
.
ignore_yum_errors=no

Like that. And if there are any errors under that section gdeploy
will exit and will not continue from there on.

https://github.com/gluster/gdeploy/commit/1ba725e795
https://github.com/gluster/gdeploy/commit/7f2e8f3ea0
https://github.com/gluster/gdeploy/commit/1acd414ba4

Are some of the commits that fixes the bug

Moving this bug back because i see that even after having ignore_yum_errors=no and ignore_register_errors=no gdeploy continues to the next step and does not exit.

Commit: https://github.com/gluster/gdeploy/commit/8a656abf5f7871 fixes the issue.

Moving this bug back as i see that in the setup-cache section i have added the line ignore_lv_errors=no and i gave a device which does not exist. I expected create logical volume fails and gdeploy just exits. But it just continues.

Commit: https://github.com/gluster/gdeploy/commit/5545054a fixes the issue.

Verified and works fine with build gdeploy-2.0.1-8.el7rhgs.noarch.

With the above feature in gdeploy i see that when ever a step fails gdeploy just exits and does not proceed further. If user wants to proceed further even if any of the section fails he needs to a add a flag called ignore_blah_errors=yes.

This feature is not completely implemented and there are still some caveats to this.

1) Run a command in shell commands does not have this feature.
2) Add / delete services does not have this feature.
3) start/stop/restart/reload services does not have this feature.
4) setup_cache section does not have this feature.
5) Running a shell script does not have this feature.
6) Update-file does not have this feature.

Will raise another bug for implementing this feature for the above listed ones.

When i say "<x> does not have this feature" what i mean here is even if the step fails gdeploy just ignores the errors and continues with the next steps and there is no option provided for user to add a flag like <ignore_blah_errors>=no so that it does not proceed.

(In reply to RamaKasturi from comment #13)
> Verified and works fine with build gdeploy-2.0.1-8.el7rhgs.noarch.
> 
> With the above feature in gdeploy i see that when ever a step fails gdeploy
> just exits and does not proceed further. If user wants to proceed further
> even if any of the section fails he needs to a add a flag called
> ignore_blah_errors=yes.
> 
> This feature is not completely implemented and there are still some caveats
> to this.
> 
> 1) Run a command in shell commands does not have this feature.
> 2) Add / delete services does not have this feature.
> 3) start/stop/restart/reload services does not have this feature.
> 4) setup_cache section does not have this feature.
> 5) Running a shell script does not have this feature.
> 6) Update-file does not have this feature.

Also [volume] section doesn't obey the generic ignore_errors=no, so one need to have 'ignore_volume_errors=no' explicitly for this section, in order to halt gdeploy execution on failure

> 
> Will raise another bug for implementing this feature for the above listed
> ones.
> 
> When i say "<x> does not have this feature" what i mean here is even if the
> step fails gdeploy just ignores the errors and continues with the next steps
> and there is no option provided for user to add a flag like
> <ignore_blah_errors>=no so that it does not proceed.

(In reply to RamaKasturi from comment #13)
> Verified and works fine with build gdeploy-2.0.1-8.el7rhgs.noarch.
> 
> With the above feature in gdeploy i see that when ever a step fails gdeploy
> just exits and does not proceed further. If user wants to proceed further
> even if any of the section fails he needs to a add a flag called
> ignore_blah_errors=yes.
> 
> This feature is not completely implemented and there are still some caveats
> to this.

You are right. Currently due to a bug it is inconsistent in a few places.
In some modules ignore_blah_errors is `yes' and in some it is `no' (`no'
is due to a bug). Though it is desirable to have no in some places, I would
like to to keep it uniform, so that it is not confusing to users.

Currently fixing the above modules to `yes'. In future releases we need
to have a discussion and come to a consensus on this requirement.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHEA-2017-0483.html