Finishes the start made in 983c9d5a53, by
handling the case where `transfer` fails for some other reason, and so the
ReadContent callback does not get run. I don't know of a case where
`transfer` does fail other than the locking dealt with in that commit, but
it's good to have a guarantee.
StoreContent and StoreContentTo had a similar problem.
Things like `getViaTmp` may decide not to run the transfer action.
And `transfer` could certianly fail, if another transfer of the same
object was in progress. (Or a different object when annex.pidlock is set.)
If the transfer action was not run, the content of the object would
not all get consumed, and so would get interpreted as protocol commands,
which would not go well.
My approach to fixing all of these things is to set a TVar only
once all the data in the transfer is known to have been read/written.
This way the internals of `transfer`, `getViaTmp` etc don't matter.
So in ReadContent, it checks if the transfer completed.
If not, as long as it didn't throw an exception, send empty and Invalid
data to the callback. On an exception the state of the protocol is unknown
so it has to raise ProtoFailureException and close the connection,
same as before.
In StoreContent, if the transfer did not complete
some portion of the DATA has been read, so the protocol is in an unknown
state and it has to close the conection as well.
(The ProtoFailureMessage used here matches the one in Annex.Transfer, which
is the most likely reason. Not ideal to duplicate it..)
StoreContent did not ever close the protocol connection before. So this is
a protocol change, but only in an exceptional circumstance, and it's not
going to break anything, because clients already need to deal with the
connection breaking at any point.
The way this new behavior looks (here origin has annex.pidlock = true so will
only accept one upload to it at a time):
git annex copy --to origin -J2
copy x (to origin...) ok
copy y (to origin...)
Lost connection (fd:25: hGetChar: end of file)
This work is supported by the NIH-funded NICEMAN (ReproNim TR&D3) project.
Fix hang when transferring the same objects to two different clients at the
same time. (Or when annex.pidlock is used, two different objects to the
same or different clients.)
Could also potentially occur if a client was downloading an object and
somehow lost connection but that git-annex-shell was still running and
holding the transfer lock.
This does not guarantee that, if `transfer` fails for some other reason,
a DATA response will be made.
This work is supported by the NIH-funded NICEMAN (ReproNim TR&D3) project.
I suspect this may be due to SQLITE_IOERR_SHORT_READ, but have not
verified.
I was able to reproduce it on Linux after running the test suite in a loop
for 1-3 hours until it failed.
The WAL mode entry change in 3963c5fcf5
may have hidden the problem I was seeing; I have not seen an ErrorIO
since then.
Cache high-resolution mtimes for improved detection of modified files in v7
(and direct mode).
Including on Windows.
With back-compat support so old low-res mtimes won't break anything, and
so the new information also won't break old versions of git-annex.
I've seen intermittent failures of the test suite with v6 for a long time,
it seems to have possibly gotten worse with the changes around v7. Or just
being unlucky; all tests failed today.
Seen on amd64 and i386 builders, repeatedly but intermittently:
unused: FAIL (4.86s)
Test.hs:928:
git diff did not show changes to unlocked file
And I think other such failures, all involving v7/v6 mode tests.
I managed to reproduce the unused failure with --keep-failures,
and inside the repo, git diff was indeed not showing any changes for
the modified unlocked file.
The two stats will be the same other than mtime; the old and new files have
the same size and inode, since the test case writes to the file and then
overwrites it.
Indeed, notice the identical timestamps:
builder@orca:~/gitbuilder/build/.t/tmprepo335$ echo 1 > foo; stat foo; echo 2 > foo; stat foo
File: foo
Size: 2 Blocks: 8 IO Block: 4096 regular file
Device: 801h/2049d Inode: 3546179 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 1000/ builder) Gid: ( 1000/ builder)
Access: 2018-10-29 22:14:10.894942036 +0000
Modify: 2018-10-29 22:14:10.894942036 +0000
Change: 2018-10-29 22:14:10.894942036 +0000
Birth: -
File: foo
Size: 2 Blocks: 8 IO Block: 4096 regular file
Device: 801h/2049d Inode: 3546179 Links: 1
Access: (0644/-rw-r--r--) Uid: ( 1000/ builder) Gid: ( 1000/ builder)
Access: 2018-10-29 22:14:10.894942036 +0000
Modify: 2018-10-29 22:14:10.898942036 +0000
Change: 2018-10-29 22:14:10.898942036 +0000
Birth: -
I'm seeing this in Linux VMs; it doesn't happen on my laptop. I've also
not experienced the intermittent test suite failures on my laptop.
So, I hope that this small delay will avoid the problem.
Update: I didn't, indeed I then reproduced the same failure on my
laptop, so it must be due to something else. But keeping this change anyway
since not needing to worry about lowish-resolution mtime in the test suite seems
worthwhile.
