Adds a dependency on filepath-bytestring, an as yet unreleased fork of
filepath that operates on RawFilePath.
Git.Repo also changed to use RawFilePath for the path to the repo.
This does eliminate some RawFilePath -> FilePath -> RawFilePath
conversions. And filepath-bytestring's </> is probably faster.
But I don't expect a major performance improvement from this.
This is mostly groundwork for making Annex.Location use RawFilePath,
which will allow for a conversion-free pipleline.
Since the sqlite branch uses blobs extensively, there are some
performance benefits, ByteStrings now get stored and retrieved w/o
conversion in some cases like in Database.Export.
Only done on those calls to getFileStatus that had a RawFilePath, not a
FilePath. The others would probably be just as fast if converted to use
it with toRawFilePath, but I'm not 100% sure.
Note that genInodeCache' uses fromRawFilePath, but that value only gets
used on Windows, so on unix the thunk will never be evaluated.
This was already optimised before, but profiling found that delEntry was
around 1.5% of the total runtime of git-annex whereis. It was being
called once per environment variable per file processed.
Fixed by better caching. Since withIndexFile is almost always run with
the same .git/annex/index file, it can cache the modified environment,
rather than re-modifying it each time called.
The parser and looking up config keys in the map should both be faster
due to using ByteString.
I had hoped this would speed up startup time, but any improvement to
that was too small to measure. Seems worth keeping though.
Note that the parser breaks up the ByteString, but a config map ends up
pointing to the config as read, which is retained in memory until every
value from it is no longer used. This can change memory usage
patterns marginally, but won't affect git-annex.
Finally builds (oh the agoncy of making it build), but still very
unmergable, only Command.Find is included and lots of stuff is badly
hacked to make it compile.
Benchmarking vs master, this git-annex find is significantly faster!
Specifically:
num files old new speedup
48500 4.77 3.73 28%
12500 1.36 1.02 66%
20 0.075 0.074 0% (so startup time is unchanged)
That's without really finishing the optimization. Things still to do:
* Eliminate all the fromRawFilePath, toRawFilePath, encodeBS,
decodeBS conversions.
* Use versions of IO actions like getFileStatus that take a RawFilePath.
* Eliminate some Data.ByteString.Lazy.toStrict, which is a slow copy.
* Use ByteString for parsing git config to speed up startup.
It's likely several of those will speed up git-annex find further.
And other commands will certianly benefit even more.
Goal is to make git-annex faster by using ByteString for all the
worktree traversal. For now, this is focusing on Command.Find,
in order to benchmark how much it helps. (All other commands are
temporarily disabled)
Currently in a very bad unbuildable in-between state.
This will speed up the common case where a Key is deserialized from
disk, but is then serialized to build eg, the path to the annex object.
Previously attempted in 4536c93bb2
and reverted in 96aba8eff7.
The problems mentioned in the latter commit are addressed now:
Read/Show of KeyData is backwards-compatible with Read/Show of Key from before
this change, so Types.Distribution will keep working.
The Eq instance is fixed.
Also, Key has smart constructors, avoiding needing to remember to update
the cached serialization.
Used git-annex benchmark:
find is 7% faster
whereis is 3% faster
get when all files are already present is 5% faster
Generally, the benchmarks are running 0.1 seconds faster per 2000 files,
on a ram disk in my laptop.
It's not necessary. And if the bare repo somehow has a pointer
file in it with the same name as a file in HEAD, that file would be
populated, which would be surprising since the file is not really under
git's control.
* git-lfs: The url provided to initremote/enableremote will now be
stored in the git-annex branch, allowing enableremote to be used without
an url. initremote --sameas can be used to add additional urls.
* git-lfs: When there's a git remote with an url that's known to be
used for git-lfs, automatically enable the special remote.
See the comment for a trace of the deadlock.
Added a new StartStage. New worker threads begin in the StartStage.
Once a thread is ready to do work, it moves away from the StartStage,
and no thread will ever transition back to it.
A thread that blocks waiting on another thread that is processing
the same key will block while in the StartStage. That other thread
will never switch back to the StartStage, and so the deadlock is avoided.
Convert Utility.Url to return Either String so the error message can be
displated in the annex monad and so captured.
(When curl is used, its errors are still not caught.)
Delete the old export dbs on upgrade.
Testing this an exporting to a directory with both exporttree=yes and
importtree=yes, it refused to let an interrupted export proceed after
upgrade, with "unsafe to overwrite file". An import resolved the
problem.
It will be populated automatically by the next command that needs data
from it, the same way it gets populated in a fresh clone. That may be a
little expensive, but it's a one time cost, and no slower than in a
fresh clone.
The old db is cleaned up when a new incremental fsck is started.
The incremental fsck won't pick up where the old one left off, but I
consider this a minor enough thing that it can just be documented and
won't be a problem.
Renamed the database to .git/annex/keysdb;
the old .git/annex/keys gets deleted during the upgrade.
It is possible that an old git-annex process is running during the
upgrade. If so, it will be able to continue using the old keys db until the
upgrade is complete, and then will presumably fail in some ugly way. Or
perhaps the upgrade will be unable to delete the open files on some
systems, and so fail with an ugly error message.
It's also possible for multiple processes to be running the upgrade
concurrently. That should be fine; they will both write the same
information into the keys db.
Other databases still need to be upgraded.
Fix bug that lost modifications to unlocked files when init is re-ran in an
already initialized repo.
In retrospect needing scanUnlockedFiles False in the direct mode upgrade
path was a good hint that it was unsafe when used with True.
However, this bug did not affect upgrade from v5. In such an upgrade, an
unlocked file that is modified is left as-is. The only place
scanUnlockedFiles True did overwrite modified unlocked files is during an
git-annex init of a repo that was already initialized by git-annex.
(I also tried a scenario where the repo had not been initialized by
git-annex yet, but was cloned from a v7 repo with an unlocked file, and the
pointer file replaced with some other content, and the data loss did not
occur in that situation.)
Since the fixed scanUnlockedFiles avoids overwriting non-pointer files,
it should be safe to run in any situation, so there's no need any longer
for the parameter.
This is a non-backwards compatable change, so not suitable for merging
w/o a annex.version bump and transition code. Not yet tested.
This improves performance of git-annex benchmark --databases
across the board by 10-25%, since eg Key roundtrips as a ByteString.
(serializeKey' produces a lazy ByteString, so there is still a
copy involved in converting it to a strict ByteString. It may be faster
to switch to using bytestring-strict-builder.)
FilePath and Key are both stored as blobs. This avoids mojibake in some
situations. It would be possible to use varchar instead, if persistent
could avoid converting that to Text, but it seems there is no good
way to do so. See doc/todo/sqlite_database_improvements.mdwn
Eliminated some ugly artifacts of using Read/Show serialization;
constructors and quoted strings are no longer stored in sqlite.
Renamed SRef to SSha to reflect that it is only ever a git sha,
not a ref name. Since it is limited to the characters in a sha,
it is not affected by mojibake, so still uses String.
When the submodule's parent repo has an adjusted unlocked branch,
it gets cloned by git, but git checks out master. git annex init then
fails because it wants to enter the adjusted branch, but:
adjusted branch adjusted/master(unlocked) already exists.
Aborting because that branch may have changes that have not yet reached master
Note that init actually then exits 0, leaving master checked out.
This could also happen, absent submodules, if the parent repo has
an adjusted unlocked branch, but it is not checked out. In the more common
case where that branch is checked out, the clone uses the same branch,
so no problem.
The choices to fix this:
* Init could delete the existing adjusted branch, and re-adjust.
But then running init inside an adjusted branch on a crippled filesystem
would lose any changes that have not been synced back to master.
* Init could sync any changes back to master, but that would be very surprising
behavior for it.
* Init could simply check out the existing adjusted branch. If the branch
is diverged from master, well, sync will sort that out later.
This mirrors the behavior of cloning a repo that has an adjusted branch
checked out that has not yet been synced back to master.
Picked this choice.
All that needs to be retained in remote.log is the sameas-uuid.
The rest of the config is eliminated. This doesn't save enough space to
bother with, but it prevents anything sensitive in the config of the
dead sameas remote from lingering around.
Note that minimizesameasdead does not update the VectorClock when
changing the log line. That's normally a no-no, but in this case,
it makes each DropDead result in the exact same file contents,
and vector clocks are not needed because the transition breaks
the history chain.
forget --drop-dead: Remove several classes of git-annex log files when they
become empty, further reducing the size of the git-annex branch.
Noticed while testing sameas uuid removal, but it could happen other times
too.
An empty log file is always treated by git-annex the same as no file
being present, and when the files are per-key, it can be a sizable space
saving to exclude them from the tree.
It would have been a lot less round-about to just make git annex dead
also add the uuids of sameas remotes to the trust.log as dead.
But, that would fail in the case where there's an unmerged other clone
that has a sameas remote that the current repo does not know about.
Then it would not get marked as dead.
Handling it at transition time avoids that scenario.
Note that the generation of trustmap' in dropDead should only
happen once, due to the partial application.
This solves the problem of sameas remotes trampling over per-remote
state. Used for:
* per-remote state, of course
* per-remote metadata, also of course
* per-remote content identifiers, because two remote implementations
could in theory generate the same content identifier for two different
peices of content
While chunk logs are per-remote data, they don't use this, because the
number and size of chunks stored is a common property across sameas
remotes.
External special remote had a complication, where it was theoretically
possible for a remote to send SETSTATE or GETSTATE during INITREMOTE or
EXPORTSUPPORTED. Since the uuid of the remote is typically generate in
Remote.setup, it would only be possible to pass a Maybe
RemoteStateHandle into it, and it would otherwise have to construct its
own. Rather than go that route, I decided to send an ERROR in this case.
It seems unlikely that any existing external special remote will be
affected. They would have to make up a git-annex key, and set state for
some reason during INITREMOTE. I can imagine such a hack, but it doesn't
seem worth complicating the code in such an ugly way to support it.
Unfortunately, both TestRemote and Annex.Import needed the Remote
to have a new field added that holds its RemoteStateHandle.
I found a way to avoid inheritance complicating anything outside of
Logs.Remote. It seems fine to require all inherited values to be
inherited and not set in the sameas remote's config. Since inherited
values will be used for stuff like encryption and perhaps chunking, which
control the actual content stored on the remote, it seems likely that
there will not be any reason to need them to vary between two remotes
that access the same underlying data store.
The newer version of containers is free; the minimum ghc version is
bundled with a newer version than that.
Initremote sets that, so after both initremote and enableremote,
the git config will be set.
Any remote that does not use Annex.SpecialRemote won't set
annex-config-uuid. But that's only Remote.Git, which doesn't use
RemoteConfig anyway.
initremote --sameas=remotename sets sameas-name and sameas-uuid
Using sameas-name rather than name prevents old git-annex initremote
from enabling a sameas remote by name, since it would not handle it
correctly.
When dropping an unlocked file, preserve its mtime, which avoids git status
unncessarily running the clean filter on the file.
If the index file has close to the same mtime as a work tree file, git will
not trust the index to be up-to-date, and re-runs the clean filter
unncessarily. Preserving the mtime when depopulating a pointer file avoids
git status doing a little (or maybe a lot) of unncessary work.
There are other places that the mtime could be preserved, including other
places where pointer files are written perhaps, but also
populatePointerFile. But, I don't know of cases where those lead to git
status doing unncessary work, so I just fixed the one I'm aware of for now.
Fix bug in handling of annex.largefiles that use largerthan/smallerthan.
When adding a modified file, it incorrectly used the file size of the old
version of the file, not the current size.
That was the only largefiles limit that didn't directly look at the file on
disk already. Added a new type to keep straight the two different ways such
a limit can be matched. I kind of wanted to extend MatchingFile or FileInfo
to indicate that the matcher is supposed to operate on files from disk or
annex, but it turned out to be too complex to implement it that way.
This also changes the LimitAnnexFiles case when lookupFileKey does not find
a key. It used to fall back to statting the file, now it always returns
False. I doubt the old code could really get to that point, but if it
somehow does, it's better for preferred content matching to be consistent.
debian oldoldstable has 2.1, and that's what i386ancient uses. It would be
better to require git 2.2, which is needed to use adjusted branches, but
can't do that w/o losing support for some old linux kernels or a
complicated git backport.
This brings back .git/annex/misctmp, but only for init. If an init
is interrupted while probing using that temp directory, the files it left
will get deleted 1 week later by a subsequent git-annex run.
Can be set to false to prevent any automatic repository upgrades.
Also, removed direct mode specific upgrade code in Annex.Init, and made
needsUpgrade always include the name/path of the repo, so if
there's a problem it's clear what repo has the problem.
And, made needsUpgrade catch any exceptions that might occur during the
upgrade, so it can display a more useful error message than just the
exception.
No longer used. The only possible user of it would be code in
Upgrade.V5, so I verified that the parts of Annex.Content it used were
not used to manipulate direct mode files.
Three reasons:
* Committing as part of an upgrade is very unusual and unexpected.
* The commit was failing with a weird error message when done during an
automatic upgrade.
* Let me remove more of that sweet^Whorrible direct mode code.
* Automatically convert direct mode repositories to v7 with adjusted
unlocked branches and set annex.thin.
* init: When run on a crippled filesystem with --version=5,
will error out, since version 7 is needed for adjusted unlocked branch.
* direct: This command always errors out as direct mode is no longer
supported.
* indirect: This command has become a deprecated noop.
* proxy: This command is deprecated because it was only needed in direct
mode. (But it continues to work.)
Also removed mentions of direct mode throughough the documentation.
I have not removed all the direct mode code yet.
When upgrading a direct mode repo to v7 with adjusted unlocked branches,
fix a bug that prevented annex.thin from taking effect for the files in
working tree.
The hard links used to be ok, but commit 8e22114735 accidentially
broke them. It repopulates the worktree file, which is already a hard link,
and when it's creating the new file, the link count is already 2, and so it
doesn't make a hard link then.
Rather than direct mode, which this is a small step on the path to
removing.
Init on a crippled filesystem already used v7 adjusted branches,
and like that, this doesn't pose any interoperability issues with old
versions of git-annex that clone the same repo, because files are only
unlocked on the adjusted branch.
Turns out that 7be690f326 broke the
test suite on the i386ancient builder. There, git show-ref --verify HEAD
fails with "'HEAD' - not a valid ref". Apparently git 2.1.4 didn't
support that.
headExists works there and does the same thing.
This code was already AGPL, except for the bit split out
to Utility/MD5.hs in commit 426053cb6c.
That commit accidentially updated the license of this file from AGPL
to GPL.
Thanks to Sean Whitton for spotting this.
In 40ecf58d4b I changed the license of code I
wrote from GPL to AGPL. But, two files containing code I wrote combined
with code by others were updated to say their license is AGPL, while in
fact part of it was (the code I wrote) but part remained under the original
license (the code written by others).
Remote/Ddar.hs is now changed entirely back to GPL 3.
Annex/DirHashes.hs stays AGPL, but I broke out Utility/MD5.hs with the code
not written by me, and corrected its license statement to GPL-2, which
is the actual version of the GPL included with the code in its original
distribution at http://www.cs.ox.ac.uk/people/ian.lynagh/md5/
I made some improvements to its API after splitting it out of git-annex,
so merge those back in.
This is groundwork for removing the embedded copy of it and depending on
it.
Also moved the managerResponseTimeout disabling to Annex.Url as it's
git-annex specific.
This commit was sponsored by Ethan Aubin on Patreon.
Support running v7 upgrade in a repo where there is no branch checked out,
but HEAD is set directly to some other ref.
This commit was sponsored by Jack Hill on Patreon.
Drop support for building with ghc older than 8.4.4, and with older
versions of serveral haskell libraries than will be included in Debian 10.
The only remaining version ifdefs in the entire code base are now a couple
for aws!
This commit should only be merged after the Debian 10 release.
And perhaps it will need to wait longer than that; it would make
backporting new versions of git-annex to Debian 9 (stretch) which
has been actively happening as recently as this year.
This commit was sponsored by Ilya Shlyakhter.
No behavior changes, but this shows everywhere that a progress meter
could be displayed when hashing a file to add to the annex.
Many of the places don't make sense to display a progress meter though,
eg when importing the copy of the file probably swamps the hashing of
the file.
init: Fix a reversion in the last release that prevented automatically
generating and setting a description for the repository.
Seemed best to factor out uuidDescMapRaw that does not
have the default mempty descrition behavior.
I don't much like that behavior, but I know things depend on it.
One thing in particular is `git annex info` which lists the uuids and
descriptions; if the current repo has been initialized in some way that
means it does not have a description, it would not show up w/o that.
(Not only repos created due to this bug might lack that. For example a repo
that was marked dead and had --drop-dead delete its git-annex branch info,
and then came back from the dead would similarly not be in the uuid.log.
Also there have been other versions of git-annex that didn't set a default
description; for years there was no default description.)
Avoid an unncessary STM transaction. This will happen when the worker
pool is not completely full of the new stage, which is the common case.
In the uncommon case, this adds only a tiny bit of overhead for the
extra traversal of the worker pool. And the thread is going to block
for some time anyway.
When all worker threads are running and enteringStage is called,
it waits for an idle slot. If all off the other threads then call it in
turn, a deadlock occurrs.
This is the same problem I didn't actually fix in
5a9842d7ed.
Fixed by doing two separate STM transactions, the first replaces its
active thread with an idle thread, and the second waits for another idle
thread. That guarantees there will eventually be an idle thread to find.
The changes to WorkerPool were necessary because it can't add an idle
thread containing the Annex state and go on to run an action using that
same state, so I had to remove the Annex state from IdleWorker.
This means that Command.Move and Command.Get don't need to
manually set the stage, and is a lot cleaner conceptually.
Also, this makes Command.Sync.syncFile use the worker pool better.
In the scenario where it first downloads content and then uploads it to
some other remotes, it will start in TransferStage, then enter VerifyStage
and then go back to TransferStage for each transfer to the remotes.
Before, it entered CleanupStage after the download, and stayed in it for
the upload, so too many transfer jobs could run at the same time.
Note that, in Remote.Git, it uses runTransfer and also verifyKeyContent
inside onLocal. That has a Annex state for the remote, with no worker pool.
So the resulting calls to enteringStage won't block in there.
While Remote.Git.copyToRemote does do checksum verification, I
realized that should not use a verification slot in the WorkerPool
to do it. Because, it's reading back from eg, a removable disk to checksum.
That will contend with other writes to that disk. It's best to treat
that checksum verification as just part of the transer. So, removed the todo
item about that, as there's nothing needing to be done.
Rather than limiting it to PerformStage and CleanupStage, this opens it
up so any number of stages can be added as needed by commands.
Each concurrent command has a set of stages that it uses, and only
transitions between those can block waiting for a free slot in the
worker pool. Calling enteringStage for some other stage does not block,
and has very little overhead.
Note that while before the Annex state was duplicated on the first call
to commandAction, this now happens earlier, in startConcurrency.
That means that seek stage actions should that use startConcurrency
and then modify Annex state won't modify the state of worker threads
they then start. I audited all of them, and only Command.Seek
did so; prepMerge changes the working directory and so has to come
before startConcurrency.
Also, the remote list is built before duplicating the state, which means
that it gets built earlier now than it used to. This would only have an
effect of making commands that end up not needing to perform any actions
unncessary build the remote list (only when they're run with concurrency
enable), but that's a minor overhead compared to commands seeking
through the work tree and determining they don't need to do anything.
The hoped for optimisation of CommandStart with -J did not materialize.
In fact, not runnign CommandStart in parallel is slower than -J3.
So, CommandStart are still run in parallel.
(The actual bad performance I've been seeing with -J in my big repo
has to do with building the remoteList.)
But, this is still progress toward making -J faster, because it gets rid
of the onlyActionOn roadblock in the way of making CommandCleanup jobs
run separate from CommandPerform jobs.
Added OnlyActionOn constructor for ActionItem which fixes the
onlyActionOn breakage in the last commit.
Made CustomOutput include an ActionItem, so even things using it can
specify OnlyActionOn.
In Command.Move and Command.Sync, there were CommandStarts that used
includeCommandAction, so output messages, which is no longer allowed.
Fixed by using startingCustomOutput, but that's still not quite right,
since it prevents message display for the includeCommandAction run
inside it too.
The goal is to be able to run CommandStart in the main thread when -J is
used, rather than unncessarily passing it off to a worker thread, which
incurs overhead that is signficant when the CommandStart is going to
quickly decide to stop.
To do that, the message it displays needs to be displayed in the worker
thread, after the CommandStart has run.
Also, the change will mean that CommandStart will no longer necessarily
run with the same Annex state as CommandPerform. While its docs already
said it should avoid modifying Annex state, I audited all the
CommandStart code as part of the conversion. (Note that CommandSeek
already sometimes runs with a different Annex state, and that has not been
a source of any problems, so I am not too worried that this change will
lead to breakage going forward.)
The only modification of Annex state I found was it calling
allowMessages in some Commands that default to noMessages. Dealt with
that by adding a startCustomOutput and a startingUsualMessages.
This lets a command start with noMessages and then select the output it
wants for each CommandStart.
One bit of breakage: onlyActionOn has been removed from commands that used it.
The plan is that, since a StartMessage contains an ActionItem,
when a Key can be extracted from that, the parallel job runner can
run onlyActionOn' automatically. Then commands won't need to worry about
this detail. Future work.
Otherwise, this was a fairly straightforward process of making each
CommandStart compile again. Hopefully other behavior changes were mostly
avoided.
In a few cases, a command had a CommandStart that called a CommandPerform
that then called showStart multiple times. I have collapsed those
down to a single start action. The main command to perhaps suffer from it
is Command.Direct, which used to show a start for each file, and no
longer does.
Another minor behavior change is that some commands used showStart
before, but had an associated file and a Key available, so were changed
to ShowStart with an ActionItemAssociatedFile. That will not change the
normal output or behavior, but --json output will now include the key.
This should not break it for anyone using a real json parser.
