Refs. swh/infra/sysadm-environment#5311

It happens that once in a while a process currently drops dead when reading
there. That should caught the issue and try again without stopping the overall
process.

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#5260

so as to allow loading the set of already moved object ids from multiple
files.

Refs. swh/infra/sysadm-environment#5260

Refs. swh/infra/sysadm-environment#5260

It catches the case, writes the object representation to a dedicated tree representation
on disk, then continues with the other objects.

Prior to this, the journal client crashed.

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Related to swh/infra/sysadm-environment#5301

Refs. swh/infra/sysadm-environment#5707

For folder with a high number of objects on disk, this allows to clean up folders along
the way instead of waiting for the full listing to be done.

Refs. swh/infra/sysadm-environment#5707

Refs. swh/infra/sysadm-environment#5707

Refs. swh/infra/sysadm-environment#5707

Refs. swh/infra/sysadm-environment#5707

Refs. swh/infra/sysadm-environment#5275

Due to previous bug fix, we may still have old erratic representation. So try not to
break the analyze tool about it.

Refs. swh/infra/sysadm-environment#4707

So we avoid creating too much listing files when we have lots of problematic files.

Refs. swh/infra/sysadm-environment#4707

There is a class in the ObjectType import which is defined by both modules.

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Check for the existence of the equivalent correct model object.

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Previous dead code was only comparing correctly revision and release objects.

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Previously, this only retrieved the content per sha1. Now it checks all hashes.
If any retrieval fails, this considers the content not found.

Refs. swh/infra/sysadm-environment#4707

When a content is not found, the comparison happens between an iterable and a
content (which is not possible). This fixes that case and improves the readability of
the actual lookup.

```
attr.exceptions.NotAnAttrsClassError: <class 'list'> is not an attrs-decorated class.
```

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

Content model comparison without ctime. This is returned as None by the cassandra
backend [1].

This is set from the Content.from_dict method if present though (and it is present in
journal entries).

[1] https://gitlab.softwareheritage.org/swh/devel/swh-storage/-/blob/master/swh/storage/cassandra/storage.py?ref_type=heads#L415

[2] https://gitlab.softwareheritage.org/swh/devel/swh-model/-/blob/master/swh/model/model.py#L1451

Refs. swh/infra/sysadm-environment#4707

Refs. swh/infra/sysadm-environment#4707

This is required if we want to be able to flush representations on disk of what we read
in cassandra or postgresql backends. It's currently empty because once read, the
generator has no longer anything left to read (during the flush).

Refs. swh/infra/sysadm-environment#4707

RawExtrinsicMetadata: RawExtrinsicMetadata objects, authority and fetcher contains an
old metadata field. It is not serialized the same way when empty. (Empty dict in
journal representation, None in storage representation)

Refs. swh/infra/sysadm-environment#4707

```
+    For some objects (e.g. origin-visit, origin-visit-status, ...), we don't have a real
+    swhid (it's a dict representation), so we'll also write the (different than swhid)
+    unique key first. That will be helpful to do a join on the other part of the
+    filesystem to retrieve the details on the objects.
```

Refs. swh/infra/sysadm-environment#4707