view tests/test-mq-safety.t @ 40022:33eb670e2834

wireprotov2: define semantics for content redirects

When I implemented the clonebundles feature and deployed it on
hg.mozilla.org using Amazon S3 as a content server, server-side CPU
and bandwidth usage dropped off a cliff and a ton of server scaling
headaches went away pretty much the instant clients with support for
clonebundles were rolled out to Firefox CI.

An obvious takeaway from that experience was that offloading server
load to scalable file servers - potentially backed by a CDN - is a
really good idea. Another takeaway was that Mercurial's wire protocol
wasn't in a good position to support data offload generally.

In wire protocol version 1, there isn't a mechanism in the protocol to
say "grab the data from over here instead." For HTTP, we could teach
the client to follow HTTP redirects. Or we could invent a media type
that encoded redirects inline. But for SSH, we were pretty much out of
luck because that protocol wasn't very flexible.

Wire protocol version 2 offers the opportunity to do something better.

The recent generic server-side content caching layer in the wire
protocol version 2 server demonstrated that it is possible to have
drop-in caching of responses to command requests. This by itself
adds tons of value and already makes the built-in server much more
scalable. But I don't want to stop there.

The existing server-side caching implementation has a big weakness:
it requires the server to send data to the client. This means that
the Mercurial server is potentially sending gigabytes of data to
thousands of clients. This is problematic because compared to scaling
static file servers, scaling dynamic servers is *hard*.

A solution to this is to "offload" serving of content to something
that isn't the Mercurial server. By offloading content serving, you
turn the Mercurial server from a centralized monolithic service to
a distributed mostly-indexing service. Assuming high rates of content
offload, this should drastically reduce the total work performed by
the Mercurial server, both in terms of CPU and data transfer. This
will make Mercurial servers vastly easier to scale.

This commit defines the semantics for "content redirects" in wire
protocol version 2. Essentially:

* Servers advertise the set of locations a response could be served
  from.
* When making requests, clients advertise the set of locations they
  are willing to fetch content from.
* Servers can then replace the inline response with one that says
  "get the response from over here instead."

This feature - when fully implemented - will allow extending the
server-side caching layer to facilitate such things as integrating
your server-side cache with a scalable blob store (such as S3 or
a CDN) and offloading most data transfer to that external service.

This feature could also be leveraged for load balancing. e.g.
requests could come into a central server and then get redirected
to an available mirror depending on server availability or locality.
There's tons of potential :)

Differential Revision: https://phab.mercurial-scm.org/D4774

  $ echo '[extensions]' >> $HGRCPATH
  $ echo 'hgext.mq =' >> $HGRCPATH

  $ hg init repo
  $ cd repo

  $ echo foo > foo
  $ hg ci -qAm 'add a file'

  $ hg qinit

  $ hg qnew foo
  $ echo foo >> foo
  $ hg qrefresh -m 'append foo'

  $ hg qnew bar
  $ echo bar >> foo
  $ hg qrefresh -m 'append bar'

Try to operate on public mq changeset

  $ hg qpop
  popping bar
  now at: foo
  $ hg phase --public qbase
  $ echo babar >> foo
  $ hg qref
  abort: cannot qrefresh public revision
  (see 'hg help phases' for details)
  [255]
  $ hg revert -a
  reverting foo
  $ hg qpop
  abort: popping would remove a public revision
  (see 'hg help phases' for details)
  [255]
  $ hg qfold bar
  abort: cannot qrefresh public revision
  (see 'hg help phases' for details)
  [255]
  $ hg revert -a
  reverting foo

restore state for remaining test

  $ hg qpush
  applying bar
  now at: bar

try to commit on top of a patch

  $ echo quux >> foo
  $ hg ci -m 'append quux'
  abort: cannot commit over an applied mq patch
  [255]


cheat a bit...

  $ mv .hg/patches .hg/patches2
  $ hg ci -m 'append quux'
  $ mv .hg/patches2 .hg/patches


qpop/qrefresh on the wrong revision

  $ hg qpop
  abort: popping would remove a revision not managed by this patch queue
  [255]
  $ hg qpop -n patches
  using patch queue: $TESTTMP/repo/.hg/patches
  abort: popping would remove a revision not managed by this patch queue
  [255]
  $ hg qrefresh
  abort: working directory revision is not qtip
  [255]

  $ hg up -C qtip
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg qpop
  abort: popping would remove a revision not managed by this patch queue
  [255]
  $ hg qrefresh
  abort: cannot qrefresh a revision with children
  [255]
  $ hg tip --template '{rev} {desc}\n'
  3 append quux


qpush warning branchheads

  $ cd ..
  $ hg init branchy
  $ cd branchy
  $ echo q > q
  $ hg add q
  $ hg qnew -f qp
  $ hg qpop
  popping qp
  patch queue now empty
  $ echo a > a
  $ hg ci -Ama
  adding a
  $ hg up null
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg branch b
  marked working directory as branch b
  (branches are permanent and global, did you want a bookmark?)
  $ echo c > c
  $ hg ci -Amc
  adding c
  $ hg merge default
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  (branch merge, don't forget to commit)
  $ hg ci -mmerge
  $ hg up default
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg log
  changeset:   2:65309210bf4e
  branch:      b
  tag:         tip
  parent:      1:707adb4c8ae1
  parent:      0:cb9a9f314b8b
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     merge
  
  changeset:   1:707adb4c8ae1
  branch:      b
  parent:      -1:000000000000
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     c
  
  changeset:   0:cb9a9f314b8b
  user:        test
  date:        Thu Jan 01 00:00:00 1970 +0000
  summary:     a
  
  $ hg qpush
  applying qp
  now at: qp

Testing applied patches, push and --force

  $ cd ..
  $ hg init forcepush
  $ cd forcepush
  $ echo a > a
  $ hg ci -Am adda
  adding a
  $ echo a >> a
  $ hg ci -m changea
  $ hg up 0
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg branch branch
  marked working directory as branch branch
  (branches are permanent and global, did you want a bookmark?)
  $ echo b > b
  $ hg ci -Am addb
  adding b
  $ hg up 0
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg --cwd .. clone -r 0 forcepush forcepush2
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 07f494440405
  updating to branch default
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ echo a >> a
  $ hg qnew patch

Pushing applied patch with --rev without --force

  $ hg push -r . ../forcepush2
  pushing to ../forcepush2
  abort: source has mq patches applied
  [255]

Pushing applied patch with branchhash, without --force

  $ hg push ../forcepush2#default
  pushing to ../forcepush2
  abort: source has mq patches applied
  [255]

Pushing revs excluding applied patch

  $ hg push --new-branch -r 'branch(branch)' -r 2 ../forcepush2
  pushing to ../forcepush2
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files

Pushing applied patch with --force

  $ hg phase --force --secret 'mq()'
  $ hg push --force -r default ../forcepush2
  pushing to ../forcepush2
  searching for changes
  no changes found (ignored 1 secret changesets)
  [1]
  $ hg phase --draft 'mq()'
  $ hg push --force -r default ../forcepush2
  pushing to ../forcepush2
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)

  $ cd ..