[5.2] Add failed job ID when dispatching a JobFailed event

[kevinsimard]

We wrote a piece of code in our Laravel application which allows us to add exception information in a separate table when a job fails. The purpose of this is was to be able to have more knowledge on what happened to a specific failed job before rerunning it with php artisan queue:retry <id>. The new table has columns like the exception message, backtrace, data and some user information.

The problem that we were facing was that the ID from the failed_job table wasn't recoverable once the failer instance was inserting a record in the database because that function wasn't returning anything (void).

This could allow people to check and play with the failed job ID when a JobFailed event is dispatched in the application. I also added a default value of NULL to make sure that this isn't breaking the previous event class definition.

[GrahamCampbell]

int|null

[kevinsimard]

@GrahamCampbell Fixed

[taylorotwell]

Not sure it's really necessary to have the value be optional. There is now way to feasible extend this event from the core I don't think.

[taylorotwell]

So now that you have the ID how do you use that to update the table with the exception information? Or how do you link the two?

[kevinsimard]

@taylorotwell I created a second table with the following columns

id
# failed_job ID
failed_job_id
# exception name
name
# exception message
message
# exception backtrace
backtrace
# exception data
data
# creation date
created_at

I also added a foreign key between failed_job_id and failed_job.id with a onDelete('cascade') which will remove the row from my new table automatically whenever a job is retried with php artisan queue:retry <id>.

The reason why this is so important for us is because we have a lot of failed jobs and sometimes it's kind of hard to know what really happened to that job so we first started by adding the backtrace to the payload (column in the failed_job table) but we didn't wanted to add all the other columns in there so we thought adding a new table would be better and easier to filter if we ever want to retry all jobs for a certain type of exceptions.

[kevinsimard]

Do you want me to create a new PR to remove the optional value?

[taylorotwell]

Nah it’s OK I left it because SyncJobs have no ID to pass.

Regards,
Taylor

On Jun 8, 2016, at 9:33 PM, Kevin Simard notifications@github.com wrote:

Do you want me to create a new PR to remove the optional value?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub #13920 (comment), or mute the thread https://github.com/notifications/unsubscribe/AAcRfrwsNtQV_ZajCPNUgmNWZTC-rtNsks5qJ3thgaJpZM4IxSwv.