Mercurial > hg
view tests/test-run-tests.t @ 26623:5a95fe44121d
clonebundles: support for seeding clones from pre-generated bundles
Cloning can be an expensive operation for servers because the server
generates a bundle from existing repository data at request time. For
a large repository like mozilla-central, this consumes 4+ minutes
of CPU time on the server. It also results in significant network
utilization. Multiplied by hundreds or even thousands of clients and
the ensuing load can result in difficulties scaling the Mercurial server.
Despite generation of bundles being deterministic until the next
changeset is added, the generation of bundles to service a clone request
is not cached. Each clone thus performs redundant work. This is
wasteful.
This patch introduces the "clonebundles" extension and related
client-side functionality to help alleviate this deficiency. The
client-side feature is behind an experimental flag and is not enabled by
default.
It works as follows:
1) Server operator generates a bundle and makes it available on a
server (likely HTTP).
2) Server operator defines the URL of a bundle file in a
.hg/clonebundles.manifest file.
3) Client `hg clone`ing sees the server is advertising bundle URLs.
4) Client fetches and applies the advertised bundle.
5) Client performs equivalent of `hg pull` to fetch changes made since
the bundle was created.
Essentially, the server performs the expensive work of generating a
bundle once and all subsequent clones fetch a static file from
somewhere. Scaling static file serving is a much more manageable
problem than scaling a Python application like Mercurial. Assuming your
repository grows less than 1% per day, the end result is 99+% of CPU
and network load from clones is eliminated, allowing Mercurial servers
to scale more easily. Serving static files also means data can be
transferred to clients as fast as they can consume it, rather than as
fast as servers can generate it. This makes clones faster.
Mozilla has implemented similar functionality of this patch on
hg.mozilla.org using a custom extension. We are hosting bundle files in
Amazon S3 and CloudFront (a CDN) and have successfully offloaded
>1 TB/day in data transfer from hg.mozilla.org, freeing up significant
bandwidth and CPU resources. The positive impact has been stellar and
I believe it has proved its value to be included in Mercurial core. I
feel it is important for the client-side support to be enabled in core
by default because it means that clients will get faster, more reliable
clones and will enable server operators to reduce load without
requiring any client-side configuration changes (assuming clients are
up to date, of course).
The scope of this feature is narrowly and specifically tailored to
cloning, despite "serve pulls from pre-generated bundles" being a valid
and useful feature. I would eventually like for Mercurial servers to
support transferring *all* repository data via statically hosted files.
You could imagine a server that siphons all pushed data to bundle files
and instructs clients to apply a stream of bundles to reconstruct all
repository data. This feature, while useful and powerful, is
significantly more work to implement because it requires the server
component have awareness of discovery and a mapping of which changesets
are in which files. Full, clone bundles, by contrast, are much simpler.
The wire protocol command is named "clonebundles" instead of something
more generic like "staticbundles" to leave the door open for a new, more
powerful and more generic server-side component with minimal backwards
compatibility implications. The name "bundleclone" is used by Mozilla's
extension and would cause problems since there are subtle differences
in Mozilla's extension.
Mozilla's experience with this idea has taught us that some form of
"content negotiation" is required. Not all clients will support all
bundle formats or even URLs (advanced TLS requirements, etc). To ensure
the highest uptake possible, a server needs to advertise multiple
versions of bundles and clients need to be able to choose the most
appropriate from that list one. The "attributes" in each
server-advertised entry facilitate this filtering and sorting. Their
use will become apparent in subsequent patches.
Initial inspiration and credit for the idea of cloning from static files
belongs to Augie Fackler and his "lookaside clone" extension proof of
concept.
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Fri, 09 Oct 2015 11:22:01 -0700 |
parents | 39577d4520ab |
children | c281c20ad0ad |
line wrap: on
line source
This file tests the behavior of run-tests.py itself. Avoid interference from actual test env: $ unset HGTEST_JOBS $ unset HGTEST_TIMEOUT $ unset HGTEST_PORT $ unset HGTEST_SHELL Smoke test ============ $ run-tests.py $HGTEST_RUN_TESTS_PURE # Ran 0 tests, 0 skipped, 0 warned, 0 failed. a succesful test ======================= $ cat > test-success.t << EOF > $ echo babar > babar > $ echo xyzzy > never happens (?) > xyzzy > nor this (?) > EOF $ run-tests.py --with-hg=`which hg` . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. failing test ================== $ cat > test-failure.t << EOF > $ echo babar > rataxes > This is a noop statement so that > this test is still more bytes than success. > EOF >>> fh = open('test-failure-unicode.t', 'wb') >>> fh.write(u' $ echo babar\u03b1\n'.encode('utf-8')) and None >>> fh.write(u' l\u03b5\u03b5t\n'.encode('utf-8')) and None $ run-tests.py --with-hg=`which hg` --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ERROR: test-failure.t output changed !. --- $TESTTMP/test-failure-unicode.t +++ $TESTTMP/test-failure-unicode.t.err @@ -1,2 +1,2 @@ $ echo babar\xce\xb1 (esc) - l\xce\xb5\xce\xb5t (esc) + babar\xce\xb1 (esc) ERROR: test-failure-unicode.t output changed ! Failed test-failure.t: output changed Failed test-failure-unicode.t: output changed # Ran 3 tests, 0 skipped, 0 warned, 2 failed. python hash seed: * (glob) [1] test --xunit support $ run-tests.py --with-hg=`which hg` --xunit=xunit.xml --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ERROR: test-failure.t output changed !. --- $TESTTMP/test-failure-unicode.t +++ $TESTTMP/test-failure-unicode.t.err @@ -1,2 +1,2 @@ $ echo babar\xce\xb1 (esc) - l\xce\xb5\xce\xb5t (esc) + babar\xce\xb1 (esc) ERROR: test-failure-unicode.t output changed ! Failed test-failure.t: output changed Failed test-failure-unicode.t: output changed # Ran 3 tests, 0 skipped, 0 warned, 2 failed. python hash seed: * (glob) [1] $ cat xunit.xml <?xml version="1.0" encoding="utf-8"?> <testsuite errors="0" failures="2" name="run-tests" skipped="0" tests="3"> <testcase name="test-success.t" time="*"/> (glob) <testcase name="test-failure-unicode.t" time="*"> (glob) <![CDATA[--- $TESTTMP/test-failure-unicode.t +++ $TESTTMP/test-failure-unicode.t.err @@ -1,2 +1,2 @@ $ echo babar\xce\xb1 (esc) - l\xce\xb5\xce\xb5t (esc) + babar\xce\xb1 (esc) ]]> </testcase> <testcase name="test-failure.t" time="*"> (glob) <![CDATA[--- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ]]> </testcase> </testsuite> $ rm test-failure-unicode.t test for --retest ==================== $ run-tests.py --with-hg=`which hg` --retest --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ERROR: test-failure.t output changed ! Failed test-failure.t: output changed # Ran 2 tests, 1 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] Selecting Tests To Run ====================== successful $ run-tests.py --with-hg=`which hg` test-success.t . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. success w/ keyword $ run-tests.py --with-hg=`which hg` -k xyzzy . # Ran 2 tests, 1 skipped, 0 warned, 0 failed. failed $ run-tests.py --with-hg=`which hg` test-failure.t --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ERROR: test-failure.t output changed ! Failed test-failure.t: output changed # Ran 1 tests, 0 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] failure w/ keyword $ run-tests.py --with-hg=`which hg` -k rataxes --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ERROR: test-failure.t output changed ! Failed test-failure.t: output changed # Ran 2 tests, 1 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] Verify that when a process fails to start we show a useful message ================================================================== NOTE: there is currently a bug where this shows "2 failed" even though it's actually the same test being reported for failure twice. $ cat > test-serve-fail.t <<EOF > $ echo 'abort: child process failed to start blah' > EOF $ run-tests.py --with-hg=`which hg` test-serve-fail.t ERROR: test-serve-fail.t output changed ! ERROR: test-serve-fail.t output changed ! Failed test-serve-fail.t: server failed to start (HGPORT=*) (glob) Failed test-serve-fail.t: output changed # Ran 1 tests, 0 skipped, 0 warned, 2 failed. python hash seed: * (glob) [1] $ rm test-serve-fail.t Running In Debug Mode ====================== $ run-tests.py --with-hg=`which hg` --debug 2>&1 | grep -v pwd + echo *SALT* 0 0 (glob) *SALT* 0 0 (glob) + echo babar babar + echo *SALT* 4 0 (glob) *SALT* 4 0 (glob) .+ echo *SALT* 0 0 (glob) *SALT* 0 0 (glob) + echo babar babar + echo *SALT* 2 0 (glob) *SALT* 2 0 (glob) + echo xyzzy xyzzy + echo *SALT* 6 0 (glob) *SALT* 6 0 (glob) . # Ran 2 tests, 0 skipped, 0 warned, 0 failed. Parallel runs ============== (duplicate the failing test to get predictable output) $ cp test-failure.t test-failure-copy.t $ run-tests.py --with-hg=`which hg` --jobs 2 test-failure*.t -n !! Failed test-failure*.t: output changed (glob) Failed test-failure*.t: output changed (glob) # Ran 2 tests, 0 skipped, 0 warned, 2 failed. python hash seed: * (glob) [1] failures in parallel with --first should only print one failure >>> f = open('test-nothing.t', 'w') >>> f.write('foo\n' * 1024) and None >>> f.write(' $ sleep 1') and None $ run-tests.py --with-hg=`which hg` --jobs 2 --first --- $TESTTMP/test-failure*.t (glob) +++ $TESTTMP/test-failure*.t.err (glob) @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. Failed test-failure*.t: output changed (glob) Failed test-nothing.t: output changed # Ran 2 tests, 0 skipped, 0 warned, 2 failed. python hash seed: * (glob) [1] (delete the duplicated test file) $ rm test-failure-copy.t test-nothing.t Interactive run =============== (backup the failing test) $ cp test-failure.t backup Refuse the fix $ echo 'n' | run-tests.py --with-hg=`which hg` -i --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. Accept this change? [n] ERROR: test-failure.t output changed !. Failed test-failure.t: output changed # Ran 2 tests, 0 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] $ cat test-failure.t $ echo babar rataxes This is a noop statement so that this test is still more bytes than success. Interactive with custom view $ echo 'n' | run-tests.py --with-hg=`which hg` -i --view echo $TESTTMP/test-failure.t $TESTTMP/test-failure.t.err (glob) Accept this change? [n]* (glob) ERROR: test-failure.t output changed !. Failed test-failure.t: output changed # Ran 2 tests, 0 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] View the fix $ echo 'y' | run-tests.py --with-hg=`which hg` --view echo $TESTTMP/test-failure.t $TESTTMP/test-failure.t.err (glob) ERROR: test-failure.t output changed !. Failed test-failure.t: output changed # Ran 2 tests, 0 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] Accept the fix $ echo " $ echo 'saved backup bundle to \$TESTTMP/foo.hg'" >> test-failure.t $ echo " saved backup bundle to \$TESTTMP/foo.hg" >> test-failure.t $ echo " $ echo 'saved backup bundle to \$TESTTMP/foo.hg'" >> test-failure.t $ echo " saved backup bundle to \$TESTTMP/foo.hg (glob)" >> test-failure.t $ echo " $ echo 'saved backup bundle to \$TESTTMP/foo.hg'" >> test-failure.t $ echo " saved backup bundle to \$TESTTMP/*.hg (glob)" >> test-failure.t $ echo 'y' | run-tests.py --with-hg=`which hg` -i 2>&1 | \ > sed -e 's,(glob)$,&<,g' --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,9 +1,9 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. $ echo 'saved backup bundle to $TESTTMP/foo.hg' - saved backup bundle to $TESTTMP/foo.hg + saved backup bundle to $TESTTMP/foo.hg (glob)< $ echo 'saved backup bundle to $TESTTMP/foo.hg' saved backup bundle to $TESTTMP/foo.hg (glob)< $ echo 'saved backup bundle to $TESTTMP/foo.hg' Accept this change? [n] .. # Ran 2 tests, 0 skipped, 0 warned, 0 failed. $ sed -e 's,(glob)$,&<,g' test-failure.t $ echo babar babar This is a noop statement so that this test is still more bytes than success. $ echo 'saved backup bundle to $TESTTMP/foo.hg' saved backup bundle to $TESTTMP/foo.hg (glob)< $ echo 'saved backup bundle to $TESTTMP/foo.hg' saved backup bundle to $TESTTMP/foo.hg (glob)< $ echo 'saved backup bundle to $TESTTMP/foo.hg' saved backup bundle to $TESTTMP/*.hg (glob)< (reinstall) $ mv backup test-failure.t No Diff =============== $ run-tests.py --with-hg=`which hg` --nodiff !. Failed test-failure.t: output changed # Ran 2 tests, 0 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] test --tmpdir support $ run-tests.py --with-hg=`which hg` --tmpdir=$TESTTMP/keep test-success.t Keeping testtmp dir: $TESTTMP/keep/child1/test-success.t (glob) Keeping threadtmp dir: $TESTTMP/keep/child1 (glob) . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. test for --time ================== $ run-tests.py --with-hg=`which hg` test-success.t --time . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. # Producing time report start end cuser csys real Test \s*[\d\.]{5} \s*[\d\.]{5} \s*[\d\.]{5} \s*[\d\.]{5} \s*[\d\.]{5} test-success.t (re) test for --time with --job enabled ==================================== $ run-tests.py --with-hg=`which hg` test-success.t --time --jobs 2 . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. # Producing time report start end cuser csys real Test \s*[\d\.]{5} \s*[\d\.]{5} \s*[\d\.]{5} \s*[\d\.]{5} \s*[\d\.]{5} test-success.t (re) Skips ================ $ cat > test-skip.t <<EOF > $ echo xyzzy > #require false > EOF $ run-tests.py --with-hg=`which hg` --nodiff !.s Skipped test-skip.t: skipped Failed test-failure.t: output changed # Ran 2 tests, 1 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] $ run-tests.py --with-hg=`which hg` --keyword xyzzy .s Skipped test-skip.t: skipped # Ran 2 tests, 2 skipped, 0 warned, 0 failed. Skips with xml $ run-tests.py --with-hg=`which hg` --keyword xyzzy \ > --xunit=xunit.xml .s Skipped test-skip.t: skipped # Ran 2 tests, 2 skipped, 0 warned, 0 failed. $ cat xunit.xml <?xml version="1.0" encoding="utf-8"?> <testsuite errors="0" failures="0" name="run-tests" skipped="2" tests="2"> <testcase name="test-success.t" time="*"/> (glob) </testsuite> Missing skips or blacklisted skips don't count as executed: $ echo test-failure.t > blacklist $ run-tests.py --with-hg=`which hg` --blacklist=blacklist \ > test-failure.t test-bogus.t ss Skipped test-bogus.t: Doesn't exist Skipped test-failure.t: blacklisted # Ran 0 tests, 2 skipped, 0 warned, 0 failed. #if json test for --json ================== $ run-tests.py --with-hg=`which hg` --json --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. ERROR: test-failure.t output changed !.s Skipped test-skip.t: skipped Failed test-failure.t: output changed # Ran 2 tests, 1 skipped, 0 warned, 1 failed. python hash seed: * (glob) [1] $ cat report.json testreport ={ "test-failure.t": [\{] (re) "csys": "\s*[\d\.]{4,5}", ? (re) "cuser": "\s*[\d\.]{4,5}", ? (re) "end": "\s*[\d\.]{4,5}", ? (re) "result": "failure", ? (re) "start": "\s*[\d\.]{4,5}", ? (re) "time": "\s*[\d\.]{4,5}" (re) }, ? (re) "test-skip.t": { "csys": "\s*[\d\.]{4,5}", ? (re) "cuser": "\s*[\d\.]{4,5}", ? (re) "end": "\s*[\d\.]{4,5}", ? (re) "result": "skip", ? (re) "start": "\s*[\d\.]{4,5}", ? (re) "time": "\s*[\d\.]{4,5}" (re) }, ? (re) "test-success.t": [\{] (re) "csys": "\s*[\d\.]{4,5}", ? (re) "cuser": "\s*[\d\.]{4,5}", ? (re) "end": "\s*[\d\.]{4,5}", ? (re) "result": "success", ? (re) "start": "\s*[\d\.]{4,5}", ? (re) "time": "\s*[\d\.]{4,5}" (re) } } (no-eol) Test that failed test accepted through interactive are properly reported: $ cp test-failure.t backup $ echo y | run-tests.py --with-hg=`which hg` --json -i --- $TESTTMP/test-failure.t +++ $TESTTMP/test-failure.t.err @@ -1,4 +1,4 @@ $ echo babar - rataxes + babar This is a noop statement so that this test is still more bytes than success. Accept this change? [n] ..s Skipped test-skip.t: skipped # Ran 2 tests, 1 skipped, 0 warned, 0 failed. $ cat report.json testreport ={ "test-failure.t": [\{] (re) "csys": "\s*[\d\.]{4,5}", ? (re) "cuser": "\s*[\d\.]{4,5}", ? (re) "end": "\s*[\d\.]{4,5}", ? (re) "result": "success", ? (re) "start": "\s*[\d\.]{4,5}", ? (re) "time": "\s*[\d\.]{4,5}" (re) }, ? (re) "test-skip.t": { "csys": "\s*[\d\.]{4,5}", ? (re) "cuser": "\s*[\d\.]{4,5}", ? (re) "end": "\s*[\d\.]{4,5}", ? (re) "result": "skip", ? (re) "start": "\s*[\d\.]{4,5}", ? (re) "time": "\s*[\d\.]{4,5}" (re) }, ? (re) "test-success.t": [\{] (re) "csys": "\s*[\d\.]{4,5}", ? (re) "cuser": "\s*[\d\.]{4,5}", ? (re) "end": "\s*[\d\.]{4,5}", ? (re) "result": "success", ? (re) "start": "\s*[\d\.]{4,5}", ? (re) "time": "\s*[\d\.]{4,5}" (re) } } (no-eol) $ mv backup test-failure.t #endif backslash on end of line with glob matching is handled properly $ cat > test-glob-backslash.t << EOF > $ echo 'foo bar \\' > foo * \ (glob) > EOF $ run-tests.py --with-hg=`which hg` test-glob-backslash.t . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. $ rm -f test-glob-backslash.t Test reusability for third party tools ====================================== $ mkdir "$TESTTMP"/anothertests $ cd "$TESTTMP"/anothertests test that `run-tests.py` can execute hghave, even if it runs not in Mercurial source tree. $ cat > test-hghave.t <<EOF > #require true > $ echo foo > foo > EOF $ run-tests.py $HGTEST_RUN_TESTS_PURE test-hghave.t . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. test that RUNTESTDIR refers the directory, in which `run-tests.py` now running is placed. $ cat > test-runtestdir.t <<EOF > - $TESTDIR, in which test-run-tests.t is placed > - \$TESTDIR, in which test-runtestdir.t is placed (expanded at runtime) > - \$RUNTESTDIR, in which run-tests.py is placed (expanded at runtime) > > $ test "\$TESTDIR" = "$TESTTMP"/anothertests > $ test "\$RUNTESTDIR" = "$TESTDIR" > $ head -n 3 "\$RUNTESTDIR"/../contrib/check-code.py > #!/usr/bin/env python > # > # check-code - a style and portability checker for Mercurial > EOF $ run-tests.py $HGTEST_RUN_TESTS_PURE test-runtestdir.t . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. #if execbit test that TESTDIR is referred in PATH $ cat > custom-command.sh <<EOF > #!/bin/sh > echo "hello world" > EOF $ chmod +x custom-command.sh $ cat > test-testdir-path.t <<EOF > $ custom-command.sh > hello world > EOF $ run-tests.py $HGTEST_RUN_TESTS_PURE test-testdir-path.t . # Ran 1 tests, 0 skipped, 0 warned, 0 failed. #endif test support for --allow-slow-tests $ cat > test-very-slow-test.t <<EOF > #require slow > $ echo pass > pass > EOF $ run-tests.py $HGTEST_RUN_TESTS_PURE test-very-slow-test.t s Skipped test-very-slow-test.t: skipped # Ran 0 tests, 1 skipped, 0 warned, 0 failed. $ run-tests.py $HGTEST_RUN_TESTS_PURE --allow-slow-tests test-very-slow-test.t . # Ran 1 tests, 0 skipped, 0 warned, 0 failed.