Previously when a sync failed, we'd retry very soon (30s) no matter how
often a sync had failed before. After this change we'll retry twice and
then back off to the regular 5min interval.
In e5269a4151 I moved the reset calls to
before the first bindValue() because we didn't consistently reset()
after exec().
However, some queries may keep a lock on the database so it is necessary
to reset() as soon as the data is read. This commit reintroduces these
reset() calls.
There also was a bug in syncjournaldb: in some error conditions the
sql query wasn't reset properly. We now always reset a query before
calling bindValue.
* If a 416 is returned and we used a Range header, try again
from scratch.
* The direct URL logic was also inconsistent for resumed downloads:
it sent the Range header but didn't check the returned
Content-Range header correctly. Now resuming is disabled for
direct URL downloads.
If lastEtag was null, a force sync was triggered. Force syncs
don't update the etag. The etag retrieval job would only run if
lastEtag was not null. So it could never become non-null.
Use the right check to determine whether a file has a blacklist entry,
SyncFileItem::FileIgnored was incorrect because that denotes files from
the ignore list or blacklisted files with no retries left.
The blacklistedInDb flag does the right thing. Rename it to
hasBlacklistEntry to be more explicit.
In some owncloud server setups multiple concurrent connections for the
same session are not supported: owncloud/core#11153
This causes issues with multiple uploads and downloads. A usual symptom
is the quota job failing and the sync aborting.
This workaround lets activity on the propagator's GET and PUT jobs
reset the timeout of all network jobs. That way, queries like the quota
job would not time out while a large up/download is in progress.
The real bug in #2219 actually was that the gui allowed users to
choose to not sync the root folder. That's not supported and no
longer possible in the UI.
Instead you may now sync a folder's files without syncing its sub-
folders.
See #2167 and #2169.
* csync_detect_update: Reupload a file if the sizes differ, similar
to how we reupload if the mtimes differ.
* PropagateUploadFileQNAM::slotPutFinished(): Abort an upload if the
size changed, similar to how we abort an upload if the mtimes changed.
* Downloadinfo entries for files that no longer need to be downloaded
are useless and can be removed. In particular, the temporary files
holding partially retrieved files are now deleted when no longer
necessary.
* The same is true for blacklist entries for paths that are no longer
being discovered.
* Same for uploadinfos for files that no longer need to be uploaded.
On signout, syncs that are running in folders are aborted. That used to
also switch the state of these folders to 'paused'. To fix that, aborts
that are caused by user actions no longer change the folder paused
state.
Aborts due to errors should still pause folders! Otherwise we'd try
resyncing them even though there was an error.
Syncing without the system exclude file would cause lots of trouble.
For instance we would try uploading the temporary files we store
partially completed downloads in.
Christian Kamm