lfs: fix the stall and corruption issue when concurrently uploading blobs
We've avoided the issue up to this point by gating worker usage with an
experimental config. See 10e62d5efa73, and the thread linked there for some of
the initial diagnosis, but essentially some data was being read from the blob
before an error occurred and `keepalive` retried, but didn't rewind the file
pointer. So the leading data was lost from the blob on the server, and the
connection stalled, trying to send more data than available.
In trying to recreate this, I was unable to do so uploading from Windows to
CentOS 7. But it reproduced every time going from CentOS 7 to another CentOS 7
over https.
I found recent fixes in the FaceBook repo to address this[1][2]. The commit
message for the first is:
The KeepAlive HTTP implementation is bugged in it's retry logic, it supports
reading from a file pointer, but doesn't support rewinding of the seek cursor
when it performs a retry. So it can happen that an upload fails for whatever
reason and will then 'hang' on the retry event.
The sequence of events that get triggered are:
- Upload file A, goes OK. Keep-Alive caches connection.
- Upload file B, fails due to (for example) failing Keep-Alive, but LFS file
pointer has been consumed for the upload and fd has been closed.
- Retry for file B starts, sets the Content-Length properly to the expected
file size, but since file pointer has been consumed no data will be uploaded,
causing the server to wait for the uploaded data until either client or
server reaches a timeout, making it seem as our mercurial process hangs.
This is just a stop-gap measure to prevent this behavior from blocking Mercurial
(LFS has retry logic). A proper solutions need to be build on top of this
stop-gap measure: for upload from file pointers, we should support fseek() on
the interface. Since we expect to consume the whole file always anyways, this
should be safe. This way we can seek back to the beginning on a retry.
I ported those two patches, and it works. But I see that `url._sendfile()` does
a rewind on `httpsendfile` objects[3], so maybe it's better to keep this all in
one place and avoid a second seek. We may still want the first FaceBook patch
as extra protection for this problem in general. The other two uses of
`httpsendfile` are in the wire protocol to upload bundles, and to upload
largefiles. Neither of these appear to use a worker, and I'm not sure why
workers seem to trigger this, or if this could have happened without a worker.
Since `httpsendfile` already has a `close()` method, that is dropped. That
class also explicitly says there's no `__len__` attribute, so that is removed
too. The override for `read()` is necessary to avoid the progressbar usage per
file.
[1] https://github.com/facebookexperimental/eden/commit/c350d6536d90c044c837abdd3675185644481469
[2] https://github.com/facebookexperimental/eden/commit/77f0d3fd0415e81b63e317e457af9c55c46103ee
[3] https://www.mercurial-scm.org/repo/hg/file/5.2.2/mercurial/url.py#l176
Differential Revision: https://phab.mercurial-scm.org/D7962
Verify that pending changesets are seen by pretxn* hooks but not by other
processes that access the destination repo while the hooks are running.
The hooks (python and external) both reject changesets after some think time,
during which another process runs pull. Each hook creates a file ('notify') to
indicate to the controlling process that it is running; the process removes the
file to indicate the hook can terminate.
init env vars
$ d=`pwd`
$ maxwait=20
utility to run the test - start a push in the background and run pull
$ dotest() {
> rm -f notify
> printf 'push '; hg -R child-push tip --template '{node}\n'
> hg -R child-push -q push > push.out 2>&1 &
>
> # wait for hook to create the notify file
> i=$maxwait
> while [ ! -f notify -a $i != 0 ]; do
> sleep 1
> i=`expr $i - 1`
> done
>
> # run pull
> hg -R child-pull -q pull
> rc=$?
>
> # tell hook to finish; notify should exist.
> rm notify
> wait
>
> cat push.out
> printf 'pull '; hg -R child-pull tip --template '{node}\n'
> return $rc
> }
python hook
$ cat <<EOF > reject.py
> import os
> import time
> def rejecthook(ui, repo, hooktype, node, **opts):
> ui.write(b'hook %s\\n' % repo[b'tip'].hex())
> # create the notify file so caller knows we're running
> fpath = os.path.join('$d', 'notify')
> f = open(fpath, 'w')
> f.close()
> # wait for ack - caller should delete the notify file
> i = int("$maxwait")
> while os.path.exists(fpath) and i > 0:
> time.sleep(1)
> i -= 1
> return True # reject the changesets
> EOF
external hook
$ cat <<EOF > reject.sh
> printf 'hook '; hg tip --template '{node}\\n'
> # create the notify file so caller knows we're running
> fpath=$d/notify
> touch \$fpath
> # wait for ack - caller should delete the notify file
> i=$maxwait
> while [ -f \$fpath -a \$i != 0 ]; do
> sleep 1
> i=\`expr \$i - 1\`
> done
> exit 1 # reject the changesets
> EOF
create repos
$ hg init parent
$ hg clone -q parent child-push
$ hg clone -q parent child-pull
$ echo a > child-push/a
$ hg -R child-push add child-push/a
$ hg -R child-push commit -m a -d '1000000 0'
test python hook
$ cat <<EOF > parent/.hg/hgrc
> [extensions]
> reject = $d/reject.py
> [hooks]
> pretxnchangegroup = python:reject.rejecthook
> EOF
$ dotest
push 29b62aeb769fdf78d8d9c5f28b017f76d7ef824b
hook 29b62aeb769fdf78d8d9c5f28b017f76d7ef824b
transaction abort!
rollback completed
abort: pretxnchangegroup hook failed
pull 0000000000000000000000000000000000000000
test external hook
$ cat <<EOF > parent/.hg/hgrc
> [hooks]
> pretxnchangegroup = sh $d/reject.sh
> EOF
$ dotest
push 29b62aeb769fdf78d8d9c5f28b017f76d7ef824b
hook 29b62aeb769fdf78d8d9c5f28b017f76d7ef824b
transaction abort!
rollback completed
abort: pretxnchangegroup hook exited with status 1
pull 0000000000000000000000000000000000000000
Test that pending on transaction without changegroup see the normal changegroup(
(issue4609)
$ cat <<EOF > parent/.hg/hgrc
> [hooks]
> pretxnchangegroup=
> pretxnclose = hg tip -T "tip: {node|short}\n"
> [phases]
> publishing=False
> EOF
setup
$ cd parent
$ echo a > a
$ hg add a
$ hg commit -m a
tip: cb9a9f314b8b
actual test
$ hg phase --public .
tip: cb9a9f314b8b