I'm not entirely happy with using a trailing / on a "file" entry for
transferring a treemanifest. We've discussed putting some flags on
each file header[0], but I'm unconvinced that's actually any better:
if we were going to add another feature to the cg format we'd still be
doing a version bump anyway to cg4, so I'm inclined to not spend time
coming up with a more sophisticated format until we actually know what
the next feature we want to stuff in a changegroup will be.
Test changes outside test-treemanifest.t are only due to the new CG3
bundlecap showing up in the wire protocol.
Many thanks to adgar@google.com and martinvonz@google.com for helping
me with various odd corners of the changegroup and treemanifest API.
0: It's not hard refactoring, nor is it a lot of work. I'm just
disinclined to do speculative work when it's not clear what the
customer would actually be.
Incoming was using bundle1 in all cases, as bundle1 is restricted to
changegroup1 and does not support general delta, this can lead to significant
CPU overhead if the server is using general delta storage. We now properly
request and store a bundle2 to disk.
If the server include any output or error in the bundle, they will be stored on
disk and replayed when the bundle is read. As 'hg incoming' is going to read the
bundle right away, we call that 'good' enough and go back to the bigger plan of
having general delta on by default.
This was tracked as 4864
$TESTDIR is added to the path, so this is superfluous. Also,
inconsistent use of quotes means we might have broken on tests with
paths containing spaces.
When the progress extension is not enabled, each call to 'ui.progress' used to
issue a debug message. This results is a very verbose output and often redundant
in tests. Dropping it makes tests less volatile to factor they do not meant to
test.
We had to alter the sed trick in 'test-rename-merge2.t'. Sed is used to drop all
output from a certain point and hidding the progress output remove its anchor.
So we anchor on something else.
As explained in a previous changeset, prioritizing heads too much behaves
pathologically when there are more heads than the sample size. To counter this,
we always inject exponential samples before reducing to the sample size limit.
This already show some benefit in the test themselves, but on a real-world example
this moves my discovery for push to pathologically headed repo from 45 rounds to
17 of them.
We should maybe ensure that at least 25% of the result sample is heads, but I
think the random sampling will be fine in practice.
The heads and exponential sample are going to end up in the same set
before any extra processing happens. We simplify the code by directly
updating a set with heads.
Changes in the order the set is built lead to small changes in the random
sampling output. But after double checking, I can confirm the input data to
the random sampling is consistent.
Before this changeset, the discovery protocol was too heads-centric. Heads of the
undiscovered set were always sent for discovery and any room remaining in the
sample were filled with exponential samples (and random ones if any room
remained).
This behaved extremely poorly when the number of heads exceeded the sample size,
because we keep just asking about the existence of heads, then their direct parent
and so on. As a result, the 'O(log(len(repo)))' discovery turns into a
'O(len(repo))' one. As a solution we take a random sample of the heads plus
exponential samples. This way we ensure some exponential sampling is achieved,
bringing back some logarithmic convergence of the discovery again.
This patch only applies this principle in one place. More places will be updated
in future patches.
One test is impacted because the random sample happen to be different. By
chance, it helps a bit in this case.
The randomness in the discovery protocol made this problem hard to reproduce.
The test mocks random.sample to make sure we hit the problem every time. The
set iteration order also made the output unstable ... but with the issue fixed,
it is stable.
Further digging on this issue show that the limit on the sample size used in
discovery never works for heads. Here is a quote from the code itself:
desiredlen = size - len(always)
if desiredlen <= 0:
# This could be bad if there are very many heads, all unknown to the
# server. We're counting on long request support here.
The long request support never landed and evolution make the "very many heads,
all unknown to the server" case quite common.
We implement a simple and stupid hard limit of sample size for all query. This
should prevent HTTP 414 error with the current state of the code.
The set discovery start by sending a "known" command with all local heads. When
the number of local heads is massive (eg: using hidden changesets) such request
becomes too large. This lead to 414 error over http, aborting the whole
process.
We limit the size of the sample used by the first query to fix this.
The test are impacted because they do test massive number of heads. But they do
not test it over real world http setup.
Pull would send a getbundle command where common heads were sent both as common
and head, even though there is no reason to request a common head.
The request was thus twice as big as necessary and more likely to hit HTTP
header size limits.
Instead, don't request heads that already are common.
This is fixed in bundlerepo.getremotechanges . It could perhaps also have been
fixed in discovery.findcommonincoming but that would have a bigger impact.
Many tests didn't change back from subdirectories at the end of the tests ...
and they don't have to. The missing 'cd ..' could always be added when another
test case is added to the test file.
This change do that tests (99.5%) consistently end up in $TESTDIR where they
started, thus making it simpler to extend them or move them around.
When setting up the next sample, we always add all of the heads, regardless
of the desired max sample size. But if the number of heads exceeds this
size, then we don't add any more nodes from the still undecided set.
(This is debatable per se, and I'll investigate it, but it's how we designed
it at the moment.)
The bug was that we always added the overall heads, not the heads of the
remaining undecided set. Thus, if #heads>200 (desired sample size), we
did not make progress any longer.
This means that we now discover both subset conditions (local<remote and
remote<local) in a single roundtrip without ever constructing an actual
sample (which takes a bit of client CPU).
Adds a new discovery method based on repeatedly sampling the still
undecided subset of the local node graph to determine the set of nodes
common to both the client and the server.
For small differences between client and server, it uses about the same
or slightly fewer roundtrips than the old tree-based discovery. For
larger differences, it typically reduces the number of roundtrips
drastically (from 150 to 4, for instance).
The old discovery code now lives in treediscovery.py, the new code is
in setdiscovery.py.
Still missing is a hook for extensions to contribute nodes to the
initial sample. For instance, Augie's remotebranches could contribute
the last known state of the server's heads.
Credits for the actual sampler and computing common heads instead of
bases go to Benoit Boissinot.