| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
|
|
|
|
|
| |
Caused by removal of BackupStoreRefCountDatabase::CreateForRegeneration.
It's not safe to do this in housekeeping anyway, because it might not
complete the scan, leaving an incomplete database, that might persuade
BackupStoreContext that it's OK to delete an object that still really has
references to it.
If your refcount db disappears or is damaged, you need to run a check fix.
|
| |
|
|
|
|
|
|
|
|
|
| |
Use a temporary refcount db for check instead of an in-memory vector. This
avoid the memory usage problems created by using the vector on large accounts,
but may require us to improve the efficiency of the refcount database itself
to avoid large numbers of small I/O operations. That is very doable now that
we're using a class for it.
Fix some inconsistencies and mistakes in handling reference counts and
info counters during account checks (more to come).
|
| |
|
|
|
| |
Caused by premature switch from GetNumFiles to GetNumCurrentFiles.
|
| |
|
|
|
|
| |
Fix erroneous detection of leaked file descriptors by closing syslog before
compiling the initial list of file descriptors.
|
| | |
|
| |
|
|
|
| |
Allows us to get information about the account that's not available
using the old AccountUsage command. Currently only used in tests.
|
| |
|
|
| |
Makes code easier to read.
|
| | |
|
| | |
|
| |
|
|
| |
Makes the output look a little nicer as bbackupquery dies.
|
| |
|
|
| |
For C++ style.
|
| | |
|
| |
|
|
|
|
|
| |
By definition, it will be cleared up by the OS automatically.
Also we don't need to support gcc 2.x any more. It was replaced by gcc3 in
2001, 12 years ago.
|
| |
|
|
| |
Helps with debugging based on exception messages alone.
|
| |
|
|
|
|
| |
This reverts commit b3bc7ad5be2c5b2588793aaa9fbc05e4a1c0aea5.
Should not have been applied in SVN (not just yet, anyway).
|
| |
|
|
| |
More accurate name for this variable.
|
| |
|
|
| |
Helps to find the right file when looking at a failed build output.
|
| |
|
|
| |
Make the release and debug options more similar.
|
| | |
|
| |
|
|
|
|
|
| |
This should never have been a valid command, but was accepted before, and
silently ignored because it was treated as an empty command. This obscured
bad quoting of bbackupquery command-line arguments, allowing commands to
appear to work, but not do what you expected.
|
| |
|
|
| |
This reverts commit 9d0580c58025ba94bee02cdc2aae3fe0b6ccf604.
|
| | |
|
| |
|
|
| |
Not really an API, but useful for BackupProtocolLocal2.
|
| |
|
|
| |
Otherwise we might detect the opened syslog socket as a leaked file descriptor.
|
| | |
|
| |
|
|
|
| |
Adds a nicer constructor and more helpful cleanup on QueryFinished()
to the auto-generated BackupProtocolLocal.
|
| |
|
|
|
| |
So we can call it on either a Client or a Local protocol object, without
having to know which it is, or cast it.
|
| |
|
|
|
| |
Allows us to call them from tests without shelling out, making debugging
easier.
|
| | |
|
| | |
|
| |
|
|
| |
Count them as Deleted, not as Old.
|
| |
|
|
|
| |
You can only pass a std::auto_ptr<IOStream> to SendStream now. Nothing
else will do for Clang. But you can cast a std::auto_ptr<subclass>.
|
| |
|
|
|
|
|
|
| |
Our copy of getopt.h in lib/win32 was beating the system getopt.h in the
search path, and providing different definitions, which clang picked up.
Renamed our getopt.h to bsd_getopt.h, and added a box_getopt.h to decide
whether to include this one instead of the system getopt.h.
|
| |
|
|
|
|
|
| |
It doesn't know that we call a function that exits the program
prematurely. In any case, this patch stops it from complaining.
|
| |
|
|
|
|
|
| |
I don't think this has any effect in practice, but this patch ensures
that it is initialised anyway.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
BOX_ERROR.
Remove debugging code for Sune Molgaard's issue with non-existent files.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
Fixes some cases where a directory might refer to an object that doesn't
exist, which is removed from the directory, but an object that depends on
it (a patch) isn't removed, and requires a subsequent recheck.
|
| | |
|
| |
|
|
|
|
| |
Thanks to Arnaud for submitting the patch:
"I propose a little patch to enable the SetProcessTitle function on windows system."
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
Report whether the directory still had errors after the first pass of
dir.CheckAndFix(), and also whether a file is marked as both Old and
Deleted, which means that its block usage is not accounted for anywhere.
There's possibly still a bug here where an invalid directory entry that
is itself a directory may not result in the parent directory being marked
as isModified and written back out afterwards, which may be what's
happening to Markus Grunwald (failure to correct the error with "check fix").
|
| |
|
|
|
|
|
| |
It's not safe to use an iterator after the underlying collection has
been modified. We need to restart iterating over the directory in that
case. Otherwise we could loop forever looking for an end() that we've
already passed, or start accessing unallocated memory.
|
