v0.6.0 job failures go into a continuous loop

[kdesjard]

Job failures go into a continuous loop regardless of # of retries.

Using sqlite and for the given job, the attempts column is zero.

[geofmureithi]

Will look into it and do a patch

[geofmureithi]

@kdesjard does the current 0.6 branch work for you?

[kdesjard]

No, it still continuously loops and the db row has 0 attempts:

sqlite> select attempts,max_attempts from jobs;
0|25

Does not work with or without a retry policy. With a retry policy of zero or one, I still get "attempt to subtract with overflow"

Am I supposed to call record_attempt ?

[geofmureithi]

Are you using the RetryLayer? Could you share a minimal example?

[geofmureithi]

I think I found the bug.
Ack does not try to increase the number of attempts.
Will fix this.

[geofmureithi]

Should be resolved in #374
The problem was ack did not update the retries.

[kdesjard]

With a retry policy of 0:

RetryLayer::new(RetryPolicy::retries(0));

A failed job runs many times (not sure if 25 or 26):

sqlite> select attempts,max_attempts from jobs;
26|25

A successful job runs once, but is the DB has 2:

select attempts,max_attempts from jobs;
2|25

[geofmureithi]

With a retry policy of 0:

RetryLayer::new(RetryPolicy::retries(0));

A failed job runs many times (not sure if 25 or 26):

sqlite> select attempts,max_attempts from jobs;
26|25

A successful job runs once, but is the DB has 2:

select attempts,max_attempts from jobs;
2|25

I think there is confusion here between RetryLayer and Storage retries. RetryLayer just defines the number of in memory retries but it's still based on attempts and max_attempts. You may want to either set max_atempts or return Error::Abort to prevent retries.