FBR: Manually delete the Fedora 29 compose that is failed in Bodhi's DB

[Randy Barlow]

There is a Fedora 29 Compose in Bodhi's DB that has gotten into a weird
state due to being run when the process didn't know about Fedora 29. To
fix it, I wish to use pshell on bodhi-backend01 to delete it:

$ sudo -u apache pshell /etc/bodhi/production.ini
> >> c = m.Compose.query.first()
> >> m.Session().delete(c)
> >> m.Session().commit() 
+1's?

[Stephen John Smoogen]

On Wed, 29 Aug 2018 at 15:21, Randy Barlow <bowlofeggs(a)fedoraproject.org&gt;
wrote:

>  There is a Fedora 29 Compose in Bodhi's DB that has gotten into a
> weird
>  state due to being run when the process didn't know about Fedora 29. To
>  fix it, I wish to use pshell on bodhi-backend01 to delete it:
>
>  $ sudo -u apache pshell /etc/bodhi/production.ini
>  >>> c = m.Compose.query.first()
>  >>> m.Session().delete(c)
>  >>> m.Session().commit()
>
>  +1's?
>
>  
Can you give me more information on what this will due, what could go wrong
and how to recover? If not I don't think I can give an answer.



>  _______________________________________________
>  infrastructure mailing list -- infrastructure(a)lists.fedoraproject.org
>  To unsubscribe send an email to
>  infrastructure-leave(a)lists.fedoraproject.org
>  Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
>  List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
>  List Archives:
>  https://lists.fedoraproject.org/archives/list/[e-mail]...
>  

-- 
Stephen J Smoogen.

[Randy Barlow]

On 08/29/2018 03:26 PM, Stephen John Smoogen wrote:
>  Can you give me more information on what this will due, what could
> go
>  wrong and how to recover? If not I don't think I can give an answer. 
Bodhi has a database model called a Compose, which is used to keep track
of each of its daily compose jobs (formerly called mashes):

https://github.com/fedora-infra/bodhi/blob/3.9.0/bodhi/server/models.py#L...

The code snippet I pasted will delete the first Compose object found in
the database. There is currently only one, and it's the one I want to
delete. To be fully honest, my shell snippet was a little lazy in this
way because I could have written a more formal query to target the
specific one, but since there is only one it's pretty easy to just say
"delete it!". I could do this instead actually:

> >> c = m.Compose.query.one() 
That will fail to retrieve it if there's more than one. So I'll do it
that way if I get my +1's, so I can be lazy *and* more precise at the
same time ☺

Since the backend isn't running any composes right now, I don't know of
any particular risk to deleting it. The Compose models are short-lived
objects in the database anyway, as they exist purely to track the state
of those jobs. When the jobs finish successfully, these Compose objects
are deleted. This particular one has stuck around because the previous
job failed. Unfortunately, every single update has been kicked out too,
so I'm not sure what would even happen if we resumed it (which would be
the typical thing to do now). This is why I'd rather just delete it so
bodhi-push can start a new one.

[Mohan Boddu]

Based on Randy's explanation +1

On Wed, Aug 29, 2018 at 3:37 PM Randy Barlow <bowlofeggs(a)fedoraproject.org&gt;
wrote:

>  On 08/29/2018 03:26 PM, Stephen John Smoogen wrote:
>  > Can you give me more information on what this will due, what could go
>  > wrong and how to recover? If not I don't think I can give an answer.
>
>  Bodhi has a database model called a Compose, which is used to keep track
>  of each of its daily compose jobs (formerly called mashes):
>
>
>  https://github.com/fedora-infra/bodhi/blob/3.9.0/bodhi/server/models.py#L...
>
>  The code snippet I pasted will delete the first Compose object found in
>  the database. There is currently only one, and it's the one I want to
>  delete. To be fully honest, my shell snippet was a little lazy in this
>  way because I could have written a more formal query to target the
>  specific one, but since there is only one it's pretty easy to just say
>  "delete it!". I could do this instead actually:
>
>  >>> c = m.Compose.query.one()
>
>  That will fail to retrieve it if there's more than one. So I'll do it
>  that way if I get my +1's, so I can be lazy *and* more precise at the
>  same time ☺
>
>  Since the backend isn't running any composes right now, I don't know of
>  any particular risk to deleting it. The Compose models are short-lived
>  objects in the database anyway, as they exist purely to track the state
>  of those jobs. When the jobs finish successfully, these Compose objects
>  are deleted. This particular one has stuck around because the previous
>  job failed. Unfortunately, every single update has been kicked out too,
>  so I'm not sure what would even happen if we resumed it (which would be
>  the typical thing to do now). This is why I'd rather just delete it so
>  bodhi-push can start a new one.
>
>  _______________________________________________
>  infrastructure mailing list -- infrastructure(a)lists.fedoraproject.org
>  To unsubscribe send an email to
>  infrastructure-leave(a)lists.fedoraproject.org
>  Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
>  List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
>  List Archives:
>  https://lists.fedoraproject.org/archives/list/[e-mail]...
>

[Stephen John Smoogen]

On Wed, 29 Aug 2018 at 15:37, Randy Barlow <bowlofeggs(a)fedoraproject.org&gt;
wrote:

>  On 08/29/2018 03:26 PM, Stephen John Smoogen wrote:
>  > Can you give me more information on what this will due, what could go
>  > wrong and how to recover? If not I don't think I can give an answer.
>
>  Bodhi has a database model called a Compose, which is used to keep track
>  of each of its daily compose jobs (formerly called mashes):
>
>
>  https://github.com/fedora-infra/bodhi/blob/3.9.0/bodhi/server/models.py#L...
>
>  The code snippet I pasted will delete the first Compose object found in
>  the database. There is currently only one, and it's the one I want to
>  delete. To be fully honest, my shell snippet was a little lazy in this
>  way because I could have written a more formal query to target the
>  specific one, but since there is only one it's pretty easy to just say
>  "delete it!". I could do this instead actually:
>
>  >>> c = m.Compose.query.one()
>
>  
The original is fine. Thank you for the training. I just wanted to make
sure there weren't more objects in the DB, that the query wouldn't give you
a random return from a hashtable etc.

+1


>  That will fail to retrieve it if there's more than one. So
> I'll do it
>  that way if I get my +1's, so I can be lazy *and* more precise at the
>  same time ☺
>
>  Since the backend isn't running any composes right now, I don't know of
>  any particular risk to deleting it. The Compose models are short-lived
>  objects in the database anyway, as they exist purely to track the state
>  of those jobs. When the jobs finish successfully, these Compose objects
>  are deleted. This particular one has stuck around because the previous
>  job failed. Unfortunately, every single update has been kicked out too,
>  so I'm not sure what would even happen if we resumed it (which would be
>  the typical thing to do now). This is why I'd rather just delete it so
>  bodhi-push can start a new one.
>
>  _______________________________________________
>  infrastructure mailing list -- infrastructure(a)lists.fedoraproject.org
>  To unsubscribe send an email to
>  infrastructure-leave(a)lists.fedoraproject.org
>  Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
>  List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
>  List Archives:
>  https://lists.fedoraproject.org/archives/list/[e-mail]...
>  

-- 
Stephen J Smoogen.

[Randy Barlow]

On 08/29/2018 03:44 PM, Stephen John Smoogen wrote:
>  The original is fine. Thank you for the training. I just wanted to
> make
>  sure there weren't more objects in the DB, that the query wouldn't give
>  you a random return from a hashtable etc.
>  
>  +1  
Excellent. I will make it so.

-- 
Number One

[Randy Barlow]

On 08/29/2018 03:26 PM, Stephen John Smoogen wrote:
>  Can you give me more information on what this will due, what could
> go
>  wrong and how to recover? If not I don't think I can give an answer. 
I have two +1's now, but I wanted to make sure you are satisfied too
before I take action. Shall I proceed?

[Kevin Fenzi]

On 08/29/2018 12:21 PM, Randy Barlow wrote:
>  There is a Fedora 29 Compose in Bodhi's DB that has gotten into a
> weird
>  state due to being run when the process didn't know about Fedora 29. To
>  fix it, I wish to use pshell on bodhi-backend01 to delete it:
>  
>  $ sudo -u apache pshell /etc/bodhi/production.ini
> >>> c = m.Compose.query.first()
> >>> m.Session().delete(c)
> >>> m.Session().commit()
>  
>  +1's? 
+1, we need to clear this to get a f29-updates-testing out.

kevin

[Randy Barlow]

This change has been made.