Because flush calls interrupt with writeLock held, it cannot interrupt
a write. Simplify by no longer defending against that.
Change-Id: Ib0b39b425335ff7b0ea1b1733562da5392576a15
StreamCopyThread#run consistently interrupts itself whenever it
discovers it has been interrupted by StreamCopyThread#flush while not
reading. The flushCount is not needed to avoid lost flushes.
All in-tree users of StreamCopyThread never flush. As a nice side
benefit, this avoids the expense of atomic operations that have no
purpose for those users.
Change-Id: I1afe415cd09a67f1891c3baf712a9003ad553062
Now that RepositoryCache have a time based eviction strategy, get rid
of the strategy to evict cache entries if heap memory is running low,
i.e. soft references. Main reason why time based eviction was
implemented was to offer an alternative to the unpredictable soft
references.
Relying on soft references is not working, especially in large heap. The
JVM GC will consider collecting soft references as last resort before
throwing an out of memory error. For example, an application like Gerrit
configured with a 128GB heap, GC will wait until all 128GB is filled
before collecting the soft references so the application will be
suffering long pauses caused by GC for a long time already. In other
words, you will have to restart application because it's unusable before
JVM eviction kicks in.
Keeping the SoftReference in RepositoryCache is causing more harm than
good. If you use the time based eviction (which is the default strategy)
and want to tune JVM to release soft references more aggressively, it
will release repositories from the cache even though they are not
expired which defeats the purpose of the repository cache.
Gerrit uses Lucene library which uses soft references and this is
causing a "memory leak" except if you configure JVM to release soft
references more aggressively which have the nasty side effect of
evicting non expired repositories from the cache.
Change-Id: I9940bd800464c7f007696d0ccde52ea617b2ebce
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
Work around issues with JSch not handling interrupts by
isolating the JSch interactions onto another thread.
Run write and flush on a single threaded Executor using
simple Callable operations wrapping the method calls,
waiting on the future to determine the outcome before
allowing the caller to continue.
If any operation was interrupted the state of the stream
becomes fuzzy at close time. The implementation tries to
interrupt the pending write or flush, but this is very
likely to corrupt the stream object, so exceptions are
ignored during such a dirty close.
Change-Id: I42e3ba3d8c35a2e40aad340580037ebefbb99b53
In case a value is used which isn’t a power of 2 there will be a high
chance of java.lang.ArrayIndexOutBoundsException and
org.eclipse.jgit.errors.CorruptObjectException due to a mismatching
assumption for the DfsBlockCache#blockSizeShift parameter.
Change-Id: Ib348b3704edf10b5f93a3ffab4fa6f09cbbae231
Signed-off-by: Philipp Marx <smigfu@googlemail.com>
* GC.tooManyLooseObjects() always responded true since the loop missed
to advance the iterator so it always incremented until the threshold was
exceeded.
* Also fix loop exit criterion which was off by 1.
* Add some tests.
Change-Id: I70976dfaa026efbcf3c46bd45941f37277a18e04
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Update to the same version used on Gerrit's master branch.
Change-Id: I20e4edd099a095c42f23df8cc57241efad2de2ce
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
StreamCopyThread.flush was introduced in
61645b938bc934fda3b0624c5bac1e3495634750 (Add timeouts to smart
transport clients, 2009-06-19) to support timeouts on write in JSch.
The commit message from that change explains:
JSch made a timeout on write difficult because they explicitly do
a catch for InterruptedException inside of their OutputStream. We
have to work around that by creating an additional thread that just
shuttles data between our own OutputStream and the real JSch stream.
The code that runs on that thread is structured as follows:
while (!done) {
int n = src.read(buf);
dst.write(buf, 0, n);
}
with src being a PipedInputStream representing the data to be written
to JSch. To add flush support, that change wanted to add an extra step
if (wantFlush)
dst.flush();
but to handle the case where the thread is blocked in the read() call
waiting for new input, it needs to interrupt the read. So that is how
it works: the caller runs
pipeOut.write(some data);
pipeOut.flush();
copyThread.flush();
to write some data and force it to flush by interrupting the read.
After the pipeOut.flush(), the StreamCopyThread reads the data that was
written and prepares to copy it out. If the copyThread.flush() call
interrupts the copyThread before it acquires writeLock and starts
writing, we throw away the data we just read to fulfill the flush.
Oops.
Noticed during the review of e67d59df3f
(StreamCopyThread: Do not let flush interrupt a write, 2016-11-04),
which introduced this bug.
Change-Id: I4aceb5610e1bfb251046097adf46bca54bc1d998
flush calls interrupt() to interrupt a pending read and trigger a
flush. Unfortunately that interrupt() call can also interrupt a
pending write, putting Jsch in a bad state and triggering "Short read
of block" errors. Add locking to ensure the flush only interrupts
reads as intended.
Change-Id: Ib105d9e107ae43549ced7e6da29c22ee41cde9d8
If there was a new flush() call during flush previous bytes, we need to
catch it in order to process the new bytes between the two flush()
calls instead of going to last catch IOException clause and end the
thread.
Change-Id: Ibc58a1fa97559238c13590aedbb85e482d85e465
Signed-off-by: Zhen Chen <czhen@google.com>
ObjectId is serializable, and so are its subtypes. Ensure that
serialization does not follow the hash collision chain internal to the
ObjectIdOwnerMap, otherwise completely unrelated objects may get
serialized when a RevObject is serialized.
Note that serializing a RevCommit or RevTag may serialize quite a few
objects due to the parent/object links they contain. A user has no real
control over how many objects will be written when a RevCommit is
serialized. C.f [1]. This change does not resolve that, but in any case
this internal hash collision chain link should not participate in
serialization.
[1] https://github.com/gitblit/gitblit/pull/1141
Change-Id: Ice331a9dc80a59ca360fcc04adaff8b5e750d847
Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
Enhance and use AtomicObjectOutputStream to write temporary files in
CleanFilter.
Change-Id: I28987dad18255a9067344f94b4e836cbd183e4b1
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
The CleanFilter leaked temporary files when a media file already existed
before running clean filter.
Change-Id: Ie20fce3f40d34095ce58e596d25d8d64fe0cde99
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Transferring data byte per byte is slow, running add with CleanFilter on
a 2.9MB file takes 20 seconds. Using a buffer of 8k shrinks this time to
70ms.
Change-Id: I3bc2d8c11fe6cfaffcc99dc2a00643e01ac4e9cc
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Previously, the streamFileThreshold, the threshold at which a file
would be streamed rather than loaded entirely into memory, was only
configurable on a global basis.
This commit makes this threshold configurable on a per-loader basis.
Bug: 490404
Change-Id: I492c18c3155dbf56eedda9044a61d76120fd75f9
Signed-off-by: Kevin Corcoran <kevin.corcoran@puppetlabs.com>
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
With the auto option, gc checks whether any housekeeping is required; if
not, it exits without performing any work. Some JGit commands run gc
--auto after performing operations that could create many loose objects.
Housekeeping is required if there are too many loose objects or too many
packs in the repository.
If the number of loose objects exceeds the value of the gc.auto option
jgit's GC consolidates all existing packs into a single pack (equivalent
to -A option), whereas git-core would combine all loose objects into a
single pack using repack -d -l. Setting the value of gc.auto to 0
disables automatic packing of loose objects.
If the number of packs exceeds the value of gc.autoPackLimit, then
existing packs (except those marked with a .keep file) are consolidated
into a single pack by using the -A option of repack. Setting
gc.autoPackLimit to 0 disables automatic consolidation of packs.
Like git the following jgit commands run auto gc:
- fetch
- merge
- rebase
- receive-pack
The auto gc for receive-pack can be suppressed by setting the config
option receive.autogc = false
Change-Id: I68a2a051b39ec2c53cb7c4b8f6c596ba65eeba5d
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
The new method addPaths(List<String>) allows callers to add multiple
paths without having to iterate over several calls to addPath(String).
Change-Id: I2c3746a97ead7118fb0ed5543a2c843224719031
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
Change-Id: Ie1e59c566b63d0dfac231e44e7ebd7f3f08f3e9f
Signed-off-by: Ned Twigg <ned.twigg@diffplug.com>
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
This SIOOBE happens reproducibly when trying to access
a repository containing Cygwin symlinks
Change-Id: I25f103fcc723bac7bfaaeee333a86f11627a92c7
Signed-off-by: Marc Strapetz <marc.strapetz@syntevo.com>
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
Remove the assumption that the local repository is a file based one.
Change-Id: I8f10fe7a54e9fc07f2a23d7901e52b65aa570d45
Signed-off-by: Thomas Meyer <thomas.mey@web.de>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Git barfs on these (and they don't make any sense), so we certainly
shouldn't write them.
Change-Id: I3faf8554a05f0fd147be2e63fbe55987d3f88099
Signed-off-by: David Turner <dturner@twosigma.com>
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
Allow for higher concurrency on DfsBlockCache by adding a configuration
for number of estimated concurrent requests.
Change-Id: Ia65e58ecb2c459b6d9c9697a2f715d933270f7e6
Signed-off-by: Philipp Marx <smigfu@googlemail.com>
According to FindBugs:
In each iteration, the String is converted to a StringBuffer/
StringBuilder, appended to, and converted back to a String. This
can lead to a cost quadratic in the number of iterations, as the
growing string is recopied in each iteration.
Replace string concatenation with StringBuffer.
Change-Id: I60e09f274bed6722f4e0e4d096b0f2b1b31ec1b4
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
* stable-4.5:
Config: do not add spaces before units
Change-Id: I54185f54e6d78d7aac873ee5f990f09582318857
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
Adding a space before the unit ('g', 'm', 'k) causes git to fail with
the error:
fatal: bad numeric config value
Change-Id: I57f11d3a1cdcca4549858e773af1a2a80fc0369f
Signed-off-by: David Turner <dturner@twosigma.com>
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
The 'factory' field is lazy initialized in the detect() method.
According to FindBugs:
Because the compiler or processor may reorder instructions, threads
are not guaranteed to see a completely initialized object, if the
method can be called by multiple threads.
Fix this by declaring the member as 'volatile'.
Change-Id: Ib32663bb28c9564584256e01f625b4e7875e6223
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
To avoid that people try to "fix" it.
Change-Id: Ib4b35e357e4c068a17243ebd2d57b058c54d5834
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
Upgrade to match the version used on Gerrit's master branch.
Requires a couple of modifications to make the tests work:
- Remove source_under_test parameters from java_test calls.
- Add vm_args with explicit setting of tmpdir location for http
tests. This is needed due to upstream changes in temporary
directory handling [1].
[1] https://github.com/facebook/buck/issues/946
Change-Id: I5d5dd5edc335d44b118e8587f69ba89b83fc7fbb
Signed-off-by: David Pursehouse <david.pursehouse@gmail.com>
Native git (as many other tools) interprets the environment variables
http_proxy, HTTP_PROXY, ... in a specific way. "http_proxy" has to be
lowercase while "https_proxy" can be lowercase or uppercase (means:
"HTTPS_PROXY"). Lowercase has precedence. This can be looked up in
"ENVIRONMENT" section of [1]. Teach JGit CLI to behave similar.
Additionally teach JGit not to interpret the environment variables if
the java process was explicitly started with the system properties
telling JVM which proxy to use. A call like "http_proxy=proxy1 java
-Dhttp.proxyHost=proxy2 ..." should use proxy2 as proxy.
[1] https://curl.haxx.se/docs/manpage.html
Change-Id: I2ad78f209792bf8f1285cf2f8ada8ae0c28f8e5a
* stable-4.5:
Unconditionally close repositories in RepositoryCache.clear()
Fix eviction of repositories with negative usage count
Adapt to parameter removed from
RepositoryCache.unregisterAndCloseRepository().
Change-Id: I7087667056ced401a3b3a027977f2715cd77a1c5
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Earlier we tried to close the repository before removing it from the
cache, so close only reduced refcount but didn't close it.
Now that we no longer leak usage count on purpose and the usage count is
now ignored anyway, there is no longer a need to run the removal twice.
Change-Id: I8b62cec6d8a3e88c096d1f37a1f7f5a5066c90a0
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
It is necessary to register a socket connection factory to prevent the
"http protocol is not supported" error when connecting over a proxy.
Change-Id: Iedf554acef841f52c1f2e3401ef0a0583ac5253b
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Jonathan Nieder