Summary:
On macos sshpeer flushes remote output later and it causes tests to fail. Let's
add a call to cleanup() to make sure the output is flushed.
This diff also makes remotefilelog test more robust.
Test Plan: Run tests on macos
Reviewers: #mercurial, tja, mitrandir
Reviewed By: mitrandir
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4721276
Tasks: 15389402
Signature: t1:4721276:1489669109:4ce59f4a1d224d57dbb7c1eb341c4e6a659d2e8c
Summary:
By default debugcheckbackup checks only one backup for the user.
With --all option it checks all backups for the user from all host/reporoots
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4700993
Tasks: 15389402
Signature: t1:4700993:1489548921:800d08f420acc8ef4f807ffd17b31837dbb3fe82
Summary:
Backing up very big repo can cause problems. And very often old commits are
not necessary at all. Let's add a config option to limit the number of heads
that are being backed up.
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4700965
Tasks: 15389402
Signature: t1:4700965:1489549163:4e2c121f01debd7b495486a1f3b062926873399d
Summary:
Previously we saved bookmark hash and revision number as state of
the last backup. Storing last backed up revision is error-prone. I saw
a few corrupted backups where bookmarks pointed to non-backed up
nodes. Also this approach puts more pressure on mysql bundle index.
On every backup it first deletes all existing backup bookmarks and then it
writes new bookmarks. Even if user changed one head or one bookmark
(and that is the usual case) current approach still deletes bookmarks first
and then rewrites them.
Instead let's store all backed up heads and bookmarks locally in json format.
Json was chosen because it's simple and human-readable. Then during backup
let's compare last backed up state to the current state of the repo, and send
to the server only the difference.
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4700930
Tasks: 15389402
Signature: t1:4700930:1489613249:a34369bf53e0718c22258304493dfa27b578857f
Summary:
Let's rename _getbackupstate() to _downloadbackupstate() because it actually
downloads it from the server. Also return class instead of tuple
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4713633
Tasks: 15389402
Signature: t1:4713633:1489612870:3439a6a99ef311bf96784b3b1e53d7e9b94c0713
Summary: New logging will have rotation and will separate different users and repos
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4689191
Tasks: 16474976
Signature: t1:4689191:1489172091:d304d4b19ae6cf52f86c11f442fa760b50d1fdf9
Summary:
For now these command will be used in test because
`wait_for_background_backup.py` tied to the backup state file. And every change
to this file requires change to wait_for_background_backup.py. Let's create
simple internal backup command.
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4667056
Tasks: 15389402
Signature: t1:4667056:1489024344:6c43241e42b7c418baa4e1542f303239ba887c45
Summary:
There can be multiple backup processes running at the same time. Since these
processes can be quite heavy it makes sense to limit them. Let's use lock file
to do that.
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4667020
Tasks: 15389402
Signature: t1:4667020:1489170813:f2a685b1c224c553d3ee004d89d3eeeca816e824
Summary:
Command that checks that every head and bookmark node is present in the
bundlestore
Test Plan: arc unit
Reviewers: #mercurial
Subscribers: #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4658436
Tasks: 15389402
Summary: This functionality will be used in command to check backup consistency
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4658433
Tasks: 15389402
Signature: t1:4658433:1488828061:49aa5bfeba922d617cf1aab393f5e9f598aa33a4
Summary:
This small change makes a way better user experience. Users usually have one
big repo per host, but they may have many hosts (for example, laptop and
devserver). If we ask for reporoot first then it's very likely that next
`pullbackup` invocation will fail because there will be ambigious hosts.
Let's switch the order.
Test Plan: arc unit
Reviewers: #mercurial, quark
Reviewed By: quark
Subscribers: quark, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4649946
Tasks: 15389402
Signature: t1:4649946:1488561645:330590fcf1dcd4af7fb572c5d4ccfd8a5ab78c60
Summary:
Next diff will add special option to specify whose backup to restore.
This diff does a necessary refactoring
Test Plan: arc unit
Reviewers: #mercurial, quark
Reviewed By: quark
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4650282
Tasks: 15389402
Signature: t1:4650282:1488562203:632a8c84d6d537663bd7d94ce9dfd18a4498ccfb
Summary:
This diff is part of the series to avoid downloading the same bundle
a few times.
Finally reuse the same bundlerepo.
Test Plan: arc unit
Reviewers: #mercurial
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4612094
Tasks: 15389402
Summary:
This diff is part of the series to avoid downloading the same bundle
a few times.
Now close all bundlerepos at once not one by one. This is necessary because
bundlerepos will be reused.
Test Plan: arc unit
Reviewers: #mercurial, durham
Reviewed By: durham
Subscribers: durham, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4612089
Tasks: 15389402
Signature: t1:4612089:1488501776:e9fc2863adebc69bf44fa5dcb1610612027508f6
Summary:
This diff is part of the series to avoid downloading the same bundle
a few times.
Test Plan: arc unti
Reviewers: #mercurial, mjpieters
Reviewed By: mjpieters
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4612073
Tasks: 15389402
Signature: t1:4612073:1487955965:03af1a25acc21d0b6d913a60dd5ef82fa7f033ad
Summary:
There are a couple of reasons to avoid using ui.username():
1) If config option ui.username is set on the server then it will use it
instead of the name of the user that does push
2) It prints confusing warning `no username found, using '....' instead`
3) In some cases it fails (probably because it calls socket.getfqdn() which
does network request).
Let's use simple util.getuser() function and set username to 'unknown' if
it fails.
Test Plan: arc unit
Reviewers: #mercurial, simonfar
Reviewed By: simonfar
Subscribers: azich, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4642457
Signature: t1:4642457:1488458040:5dfac435bf44dfa6ab7725d2f5800963e8c860ec
Summary:
Previously --list-remote wasn't user friendly. User have to always specify a
pattern, and specifying * just won't work because of bash globbing. Also
it wasn't possible to specify many patterns at once.
This diff allows specifying many patterns, it allows to not specify
patterns at all and also it allows to specify default scratch patterns to fetch
if no patterns were specified. This is useful if there are many auto-generated
bookmarks under the same scratch bookmark pattern.
Test Plan: arc unit
Reviewers: #mercurial, mjpieters
Reviewed By: mjpieters
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4605548
Signature: t1:4605548:1487934838:066a12c28dda16fa8f90674d736d21345631ef7e
Summary:
discovery.findcommonoutgoing() explicitly skips secret commits. Because of it
we also need to skip bookmarks that point to secret commits, otherwise they
will point to non-existent nodes.
Test Plan: arc unit
Reviewers: #mercurial, mjpieters
Reviewed By: mjpieters
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4611393
Tasks: 15389402
Signature: t1:4611393:1487933210:72e79923c944b13204f4cde64d415076703bbe47
Summary:
hasscratchbookmarks is True if scratchbookmarks is not empty and vice-versa.
No need to use this variable
Test Plan: arc unit
Reviewers: #mercurial, mjpieters
Reviewed By: mjpieters
Subscribers: mjpieters
Differential Revision: https://phabricator.intern.facebook.com/D4590917
Signature: t1:4590917:1487682597:c2a6e0accbd8e539a885dd11e6819986e57d24df
Summary:
listkeyspatterns support batching. Let's use and avoid making many network
calls to the hg server.
I had to modify tests because scratchbookmarks are passed to the listkeyspatterns,
and scratchbookmarks is dict and the order of the keys is unpredictable.
Because of it bundles from the server will be sent in unpredictable order, and log
output may be different.
Test Plan: arc unit
Reviewers: #mercurial, simonfar
Reviewed By: simonfar
Subscribers: mjpieters
Differential Revision: https://phabricator.intern.facebook.com/D4590914
Signature: t1:4590914:1487788586:dd5ba0dd41b0cd19f790755d7b3c1fc5ba4ade61
Summary:
Let's pull node during update if not found locally.
This is a part of selectivepull functionality.
See remotenames extensions for details about selectivepull.
Test Plan: arc unit
Reviewers: #sourcecontrol, durham
Reviewed By: durham
Subscribers: indragie, mjpieters, sergeyb
Differential Revision: https://phabricator.intern.facebook.com/D4536129
Tasks: 12479658
Signature: t1:4536129:1486667537:3d1df30cb5d1db0dd7451756102ccafee20789d5
Summary:
Default lock timeout value is very small (around couple of secs).
Let's set it for something bigger by default
Test Plan: arc unit
Reviewers: rmcelroy, mitrandir, durham
Reviewed By: durham
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4435236
Tasks: 15389402
Signature: t1:4435236:1484880747:d46276a82782ef551cda0165dea686f267c13ff9
Summary:
On a big repo with many old heads bundle that is sent to the server can become
big (in some cases even 500 Mb). That looks like a waste of bundlestore space
and will probably make backup and restore slower. Most of the space is taken
by manifest deltas because it prefers to diff manifest against previous commit
in the bundle. There are two possible approaches to reduce the size:
1) Send many small bundles (for example, one bundle per head)
2) Wrap deltaparent function and diff against actual parent previous commit in the bundle.
I chose the second approach for the following reasons:
1) It's easier to implement (main reason)
2) Many bundles probably means slower restore because there will be many requests to the bundlestore instead of just one
With this diff bundle size was reduced from 500 Mb to 8 Mb.
It can potentially increase CPU usage. I'm not sure how bad is it and will investigate it more.
Test Plan: Run `test-infinitepush-*`
Reviewers: rmcelroy, mitrandir, durham
Reviewed By: durham
Subscribers: simpkins, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4384053
Tasks: 12479677
Signature: t1:4384053:1483664446:62ec30fad433e8d279758926199a8330cb73ed2b
Summary:
If client pulled a commit without filelogs (because of remotefilelog) and
this commit was later stripped on the server, then attempt to backup this
commit fails because of missing filelogs. It's a rare issue but unfortunately
it happens sometimes. To fix it let's exclude from the backup all the
commits and their descendants that doesn't have all necessary filelogs.
Test Plan: Run infinitepush tests
Reviewers: mitrandir, rmcelroy, durham
Reviewed By: rmcelroy
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4365747
Signature: t1:4365747:1483466164:560cdf5cd2369cd2603dfd0fe6b30d7a70951f00
Summary:
`pushbackup` command became huge. This diff splits it into a few
smaller functions
Test Plan: Run `test-infinitepush-*`
Reviewers: rmcelroy, mitrandir, durham
Reviewed By: durham
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4384008
Tasks: 12479677
Signature: t1:4384008:1483659549:a97b63a38b702a55d19cf6b47a6fd6b2547b9168
Summary: __init__.py became huge and I suggest to split it
Test Plan: Run `test-infinitepush-*`
Reviewers: rmcelroy, mitrandir, durham
Reviewed By: durham
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4383661
Tasks: 12479677
Signature: t1:4383661:1483659134:eaf3d7e217633895a65ad568831d9c21ea0f18d8
Summary:
`pushbackup` failed when there are hidden heads in the repo but no visible
heads (see test for example). This diff fixes it.
Test Plan: Run `test-infinitepush-*`
Reviewers: durham, mitrandir, rmcelroy
Reviewed By: rmcelroy
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4365677
Tasks: 12479677
Signature: t1:4365677:1483465909:fd1d96211909d3eeccab4fd17a364af4a1997f85
Summary:
The '&>>' operator only works in linux, not OSX. So let's replace it with a more
verbose equivalent. 'man bash' says these two are equivalent.
Test Plan:
Running the tests normally on OSX don't actually catch this, because
this particular test is skipped because it doesn't see that evolve is installed.
We need to run the tests with:
PYTHONPATH=/opt/facebook/hg/lib/python2.7/site-packages/:/opt/homebrew/lib/python2.7/site-packages/ /opt/homebrew/opt/python27/bin/python2.7 ../../facebook-hg-rpms/hg-crew/tests/run-tests.py test-infinitepush-backup.t
To get coverage. And the test is now passing on OSX and linux.
Reviewers: stash, #mercurial, quark
Reviewed By: quark
Subscribers: mjpieters
Differential Revision: https://phabricator.intern.facebook.com/D4379593
Signature: t1:4379593:1483488383:ce8a045bbc29d719eafa372db56453b83f1d8df8
Summary:
Default path was used during background backup even if non-default path was
passed on the command line. So `hg push somepath --background` is
equivalent to `hg push --background`. This is not correct and this diff fixes it
Test Plan: Run `test-infinitepush-*`
Reviewers: durham, rmcelroy, mitrandir
Reviewed By: mitrandir
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4357295
Tasks: 12479677
Signature: t1:4357295:1482330125:d2b043f8035046ab43667b4387bf642e53261681
Summary:
As the name suggest it will restore backup made by `hg pushbackup`.
If user has only one backup for the `dest` repo then it will be restored.
But user may have backed up many local repos that points to `dest` repo.
These local repos may reside on different hosts or in different
repo roots. It makes restore ambiguous; `--reporoot` and `--hostname`
options are used to disambiguate.
Example situation:
1) User has only one laptop with mercurial repo `repo` that was cloned
from remote server. He or she run `hg pushbackup`. Then laptop breaks
and user gets a new one, clones the `repo` again and runs `hg restore`.
It automatically restores the backup.
2) User has devserver and laptop and backups were made from both.
Then if user decides to switch devserver and run `hg restore` on the new
devserver he or she has to specify `--hostname`.
Future plans:
1) Add `--user` option to make it possible to restore another user's backup
Test Plan: Run `test-infinitepush-*`
Reviewers: rmcelroy, mitrandir, durham
Reviewed By: durham
Subscribers: mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4280832
Tasks: 12479677
Signature: t1:4280832:1481565335:2a21ceafa2ff80242076a79693046514434afb40
Summary:
`bundlerepo` should be closed to ensure that temp file is deleted.
`bundlerepo` creates temp file when bundle is compressed. This is *not* the
case in infinitepush (all our bundle are uncompressed). But seems that
upstream cg1unpacker.compressed() returns True even if bundle is uncompressed.
Also this diff makes `_getoutputbundleraw()` read output bundle in memory
instead of returning a generator. The reason for doing it is because generator
becomes invalid as soon as `bundlerepo` is closed.
This approach will obviously increase memory usage but it shouldn't be a
problem since bundles are small (no more than 10s Mb).
#thanks @quark for reporting it
Test Plan: Run `test-infinitepush-*`
Reviewers: durham, rmcelroy, mitrandir, quark
Reviewed By: quark
Subscribers: quark, mjpieters, #sourcecontrol
Differential Revision: https://phabricator.intern.facebook.com/D4327186
Tasks: 12479677
Signature: t1:4327186:1481738559:2c6b22c305da4d572da9de21dfdf1179f7281744
