Recognize not processing the queue

Hello again everyone,

I got recognize to finally work, which is great, and it started doing it’s job like a charm, but then it stopped doing anything and hasn’t processed a single picture for more than 11 hours, and there is still a ton of picture to process in the queue.

Object recognition: 18070 Queued files, Last classification: 11 hours ago, Scheduled background jobs: 1, Last background job execution: 11 hours ago

The configuration panel for recognize seems to say that a job should start every 5 minutes.

Since after reinstalling the app it seemed some things were misconfigured, I was wondering if the job was maybe not well created.

I check the oc_job table, but it doesn’t seem to tell when will the next execution be. Is there any way to check this?

MariaDB [nextcloud]> select * from oc_jobs;
| id    | class                                                           | argument                                                     | last_run   | last_checked | reserved_at | execution_duration | argument_hash                    | time_sensitive |
| 59728 | OCA\Recognize\BackgroundJobs\ClassifyImagenetJob                | {"storageId":6,"rootId":34}                                  | 1714369836 |   1714370243 |           0 |                407 | 87764c38a544d98ae2e265de3d3520bc |              0 |
| 59765 | OCA\Recognize\BackgroundJobs\ClassifyLandmarksJob               | {"storageId":6,"rootId":34}                                  | 1714370103 |   1714370103 |           0 |                 10 | 87764c38a544d98ae2e265de3d3520bc |              0 |

Is there maybe (another) setting I missed?

Edit: I am running the latest version of Recognize (6.1.1) with Nextcloud 28.0.5 on Debian 11.9.

1 Like

having the same issue on version 29.

1 Like

It actually appears to be running once a day (or at least it has run this morning), which is less often than I expected but should process everything eventually.
Still, is this behavior expected?

The jobs should run every 5 minutes. Are there any errors in the logs related to recognize?

I have the same with NC 29 @marcelklehr and no errors in logs

| OCA\Recognize\BackgroundJobs\ClassifyImagenetJob | 2024-05-05 17:55:48 | 2024-05-05 17:55:48 |
| OCA\Recognize\BackgroundJobs\ClassifyFacesJob | 2024-05-05 17:59:18 | 2024-05-05 17:59:18 |

These jobs has run once each 500 files processed successfully

No error linked to Recognize either. But at least, after a week of daily (but still only once a day) processing, all my pictures are sorted which is nice.

Same here and I edited the job php to log when it run,

Next runs are occurred at 03:03 03:08 03:23 03:28

Job had skipped some occurrences

03:03 03:08 03:23 03:28

but that doesn’t look too bad. Some occurrences skipped is kind of expected

Yesterday only at 20h23 20h28 and nothing until 03h03

Could you try logging in nextcloud/cron.php all jobs that are executed, if any? It must be doing something in that time.

Yes the cron task is over 5 min

When recognize is running I see this

crond: user www-data: process already running: php -f /var/www/html/cron.php

So cron is over 5 minutes duration, after this, recognize never run, after few hours recognize runs again

I reduced pool job from 500 to 200 and retried, all is ok

Overtime seems to be the issue

My cpu is i5-8500T