Fixed.

Removed "h/" from all #includes.

This reverts commit c132cc69.

The aborting the session after problem with mount details log message now includes the errorMessage text

Fixed.

Fixed CASTOR-4939: The privilege (B/W) list display is broken when dealing with replication requests.

The fix is to reintroduce a name for object id = 133, formerly
StageDiskCopyReplicaRequest, and to introduce a 'DELETED TYPE' that went
missing to fix the seg fault. Also many now unused constants have been
dropped.
Also added some relevant cases to the test suite to avoid future
regressions.

Simplified repack job: Oracle actually prevents multiple instances of the same job from running concurrently, so no protection is needed. On the other side, we also (re)process STARTING requests as it has happened that they have been left behind.

The new configuration directive "xcastor2.skipabort" is by default false and
can be enabled by setting it to true. If this is enabled then we no longer
abort the in-flight requests for which the client disconncted i.e. we no longer
issue stage abort requests.

Note: This is put in place to accommodate the Alice workflow. The proper way of
achieving this on the client side would be to do a prepare request before trying
to open the file.

        be tested on the client side to determine if the file is on tape.

parameter in CastorConfig to cleanup old stuck d2d jobs.

More amendmends of the 2.1.15-0 Release Notes following production upgrades. Cf. also bug CASTOR-4932.

This reverts commit f239e684.

The tape alert detection was missing in TapeWriteSingleThread::TapeCleaning::~TapeCleaning, it was added only to
TapeReadSingleThread::TapeCleaning::~TapeCleaning.

        properly check for errors returned by the getxattr function.

Fixed missing early checks on the return values of rfio, leading to downstream (and harder to understand errors).

We now delete entirely the migration mounts when there is no tape available.
If suffucient migration jobs are still present, the migration mount creation logic will recreate a migartion mount
within 1-2 minutes. If all migration jobs are deleted, the warning will only appear once.

to the list of Oracle error codes that shall trigger a reconnection.

This error is not properly a 'lost connection' error, but we have
repeatedly observed cases where the DB connection gets corrupted
and the server spits out this error message for perfectly valid
statements (last occurrence being an incident in castorpublic this
morning). A reconnection is a valid workaround to get back in
business.

The logic from the session cleaner has been duplicated to the RAII TapeCleaning class's destructor
for both migrations and recalls.

Fixed.

Revert "Moved garbage collector's implementation to using NObjectIterator instead of the deprecated librados::ObjectIterator."

This reverts commit 3381b2de.

The old implementation of Rados against which we compile for production does not support this feature yet (but it is deprecated
in later versions of Rados).

Moved garbage collector's implementation to using NObjectIterator instead of the deprecated librados::ObjectIterator.

Mark a 0-space (unmounted) filesystem as disabled, so that if it used to contain
files, access are not attempted to them.

Fixed bad error handling when deleting an NS entry that does not exists, plus fixed some pylint warnings