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
#require svn svn-bindings
$ cat >> $HGRCPATH <<EOF
> [extensions]
> convert =
> EOF
$ svnadmin create svn-repo
$ svnadmin load -q svn-repo < "$TESTDIR/svn/move.svndump"
$ SVNREPOPATH=`pwd`/svn-repo
$ SVNREPOURL="`"$PYTHON" $TESTDIR/svnurlof.py \"$SVNREPOPATH\"`"
Convert trunk and branches
$ hg convert --datesort "$SVNREPOURL"/subproject A-hg
initializing destination A-hg repository
scanning source...
sorting...
converting...
13 createtrunk
12 moved1
11 moved1
10 moved2
9 changeb and rm d2
8 changeb and rm d2
7 moved1again
6 moved1again
5 copyfilefrompast
4 copydirfrompast
3 add d3
2 copy dir and remove subdir
1 add d4old
0 rename d4old into d4new
$ cd A-hg
$ hg log -G --template '{rev} {desc|firstline} files: {files}\n'
o 13 rename d4old into d4new files: d4new/g d4old/g
|
o 12 add d4old files: d4old/g
|
o 11 copy dir and remove subdir files: d3/d31/e d4/d31/e d4/f
|
o 10 add d3 files: d3/d31/e d3/f
|
o 9 copydirfrompast files: d2/d
|
o 8 copyfilefrompast files: d
|
o 7 moved1again files: d1/b d1/c
|
| o 6 moved1again files:
| |
o | 5 changeb and rm d2 files: d1/b d2/d
| |
| o 4 changeb and rm d2 files: b
| |
o | 3 moved2 files: d2/d
| |
o | 2 moved1 files: d1/b d1/c
| |
| o 1 moved1 files: b c
|
o 0 createtrunk files:
Check move copy records
$ hg st --rev 12:13 --copies
A d4new/g
d4old/g
R d4old/g
Check branches
$ hg branches
default 13:* (glob)
d1 6:* (glob)
$ cd ..
$ mkdir test-replace
$ cd test-replace
$ svnadmin create svn-repo
$ svnadmin load -q svn-repo < "$TESTDIR/svn/replace.svndump"
Convert files being replaced by directories
$ hg convert svn-repo hg-repo
initializing destination hg-repo repository
scanning source...
sorting...
converting...
6 initial
5 clobber symlink
4 clobber1
3 clobber2
2 adddb
1 clobberdir
0 branch
$ cd hg-repo
Manifest before
$ hg -v manifest -r 1
644 a
644 d/b
644 d2/a
644 @ dlink
644 @ dlink2
644 dlink3
Manifest after clobber1
$ hg -v manifest -r 2
644 a/b
644 d/b
644 d2/a
644 dlink/b
644 @ dlink2
644 dlink3
Manifest after clobber2
$ hg -v manifest -r 3
644 a/b
644 d/b
644 d2/a
644 dlink/b
644 @ dlink2
644 @ dlink3
Manifest after clobberdir
$ hg -v manifest -r 6
644 a/b
644 d/b
644 d2/a
644 d2/c
644 dlink/b
644 @ dlink2
644 @ dlink3
Try updating
$ hg up -qC default
$ cd ..
Test convert progress bar
$ cat >> $HGRCPATH <<EOF
> [progress]
> assume-tty = 1
> delay = 0
> changedelay = 0
> format = topic bar number
> refresh = 0
> width = 60
> EOF
$ hg convert svn-repo hg-progress
\r (no-eol) (esc)
scanning [=====> ] 1/7\r (no-eol) (esc)
scanning [===========> ] 2/7\r (no-eol) (esc)
scanning [=================> ] 3/7\r (no-eol) (esc)
scanning [========================> ] 4/7\r (no-eol) (esc)
scanning [==============================> ] 5/7\r (no-eol) (esc)
scanning [====================================> ] 6/7\r (no-eol) (esc)
scanning [===========================================>] 7/7\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [ ] 0/7\r (no-eol) (esc)
getting files [=====> ] 1/6\r (no-eol) (esc)
getting files [============> ] 2/6\r (no-eol) (esc)
getting files [==================> ] 3/6\r (no-eol) (esc)
getting files [=========================> ] 4/6\r (no-eol) (esc)
getting files [===============================> ] 5/6\r (no-eol) (esc)
getting files [======================================>] 6/6\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [=====> ] 1/7\r (no-eol) (esc)
scanning paths [ ] 0/1\r (no-eol) (esc)
getting files [======================================>] 1/1\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [===========> ] 2/7\r (no-eol) (esc)
scanning paths [ ] 0/2\r (no-eol) (esc)
scanning paths [==================> ] 1/2\r (no-eol) (esc)
getting files [========> ] 1/4\r (no-eol) (esc)
getting files [==================> ] 2/4\r (no-eol) (esc)
getting files [============================> ] 3/4\r (no-eol) (esc)
getting files [======================================>] 4/4\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [=================> ] 3/7\r (no-eol) (esc)
scanning paths [ ] 0/1\r (no-eol) (esc)
getting files [======================================>] 1/1\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [=======================> ] 4/7\r (no-eol) (esc)
scanning paths [ ] 0/1\r (no-eol) (esc)
getting files [======================================>] 1/1\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [=============================> ] 5/7\r (no-eol) (esc)
scanning paths [ ] 0/1\r (no-eol) (esc)
getting files [===> ] 1/8\r (no-eol) (esc)
getting files [========> ] 2/8\r (no-eol) (esc)
getting files [=============> ] 3/8\r (no-eol) (esc)
getting files [==================> ] 4/8\r (no-eol) (esc)
getting files [=======================> ] 5/8\r (no-eol) (esc)
getting files [============================> ] 6/8\r (no-eol) (esc)
getting files [=================================> ] 7/8\r (no-eol) (esc)
getting files [======================================>] 8/8\r (no-eol) (esc)
\r (no-eol) (esc)
\r (no-eol) (esc)
converting [===================================> ] 6/7\r (no-eol) (esc)
scanning paths [ ] 0/3\r (no-eol) (esc)
scanning paths [===========> ] 1/3\r (no-eol) (esc)
scanning paths [========================> ] 2/3\r (no-eol) (esc)
getting files [===> ] 1/8\r (no-eol) (esc)
getting files [========> ] 2/8\r (no-eol) (esc)
getting files [=============> ] 3/8\r (no-eol) (esc)
getting files [==================> ] 4/8\r (no-eol) (esc)
getting files [=======================> ] 5/8\r (no-eol) (esc)
getting files [============================> ] 6/8\r (no-eol) (esc)
getting files [=================================> ] 7/8\r (no-eol) (esc)
getting files [======================================>] 8/8\r (no-eol) (esc)
\r (no-eol) (esc)
initializing destination hg-progress repository
scanning source...
sorting...
converting...
6 initial
5 clobber symlink
4 clobber1
3 clobber2
2 adddb
1 clobberdir
0 branch
$ cd ..