| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
| |
Add endpoint for querying all chunk infos in an oplog.
|
| |
|
| |
- Improvement: Reduce memory usage in GC and diskbucket flush
|
| |
|
|
|
|
| |
- Improvement: GCv2: Use separate PreCache step to improve concurrency when checking references
- Improvement: GCv2: Improved verbose logging
- Improvement: GCv2: Sort chunks to read by block/offset when finding references
- Improvement: GCv2: Exit as soon as no more unreferenced items are left
|
| |
|
| |
previously would not shut down the threadpool properly which would leave threads around. This was not a problem in practice for our usage since we keep the thread pool alive for the duration of the process but it's better to clean up properly.
|
| |
|
| |
- Improvement: Use two global worker thread pools instead of ad-hoc creation of worker pools
|
| |
|
|
| |
previously, the millisecond part was inexplicably logged after the date, not the time.
|
| | |
|
| | |
|
| |
|
|
| |
* close thread pool at destruction
* parallell casimpl::initialize
|
| | |
|
| |
|
|
|
|
| |
- Improvement: Added more trace scopes for GCv2
- Bugfix: Make sure we can override flags to "false" when running `zen gc` commmand
- `smallobjects`, `skipcid`, `skipdelete`, `verbose`
|
| |
|
|
| |
controlled manner (#573)
|
| |
|
|
| |
* fix missing locks/sync of log position when writing index snapshots
* changelog
|
| |
|
| |
* gcv2 tests for project store and bugfixes
|
| |
|
|
|
| |
the previous implementation of in-memory index snapshots serialise data to memory before writing to disk and vice versa when reading. This leads to some memory spikes which end up pushing useful data out of system cache and also cause stalls on I/O operations.
this change moves more code to a streaming serialisation approach which scales better from a memory usage perspective and also performs much better
|
| |
|
|
|
| |
- Feature: New endpoint `/admin/gc-stop` to cancel a running garbage collect operation
- Feature: Added `zen gc-stop` command to cancel a running garbage collect operation
- Bugfix: GCv2 - make sure to discover all projects and oplogs before checking for expired data
|
| |
|
|
| |
- GcScheduler will now cancel any running GC when it shuts down.
- Old GC is rather limited in *when* it reacts to cancel of GC. GCv2 is more responsive.
|
| | |
|
| | |
|
| |
|
|
|
| |
* added --powercycle option. when this is passed in the zenserver process will shut down immediately after initialization is complete. This is primarily useful when benchmarking init/cleanup but could also be used to verify/clean up disk state
* moved EmptyStandbyList code to make it accessible to more commands
|
| | |
|
| |
|
|
|
|
| |
- Feature: Added xmake task `updatefrontend` which updates the zip file containing the frontend html (`/src/zenserver/frontend/html.zip`)
- Improvement: The frontend html content is no longer appended at the end of the executable which prevented signing, instead it is compiled in from the `/src/zenserver/frontend/html.zip` archive
- Improvement: MacOS now does ad-hoc code signing by default when issuing `xmake bundle`, signing with proper cert is done on CI builds
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
* reduce work when there are no blocks to compact
* fix lock scopes
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
- Feature: Added new options to zenserver for GC V2
- `--gc-compactblock-threshold` GCV2 - how much of a compact block should be used to skip compacting the block, default is 90%
- `--gc-verbose` GCV2 - enable more verbose output when running a GC pass
- Feature: Added new options to `zen gc` command for GC V2
- `--compactblockthreshold` GCV2 - how much of a compact block should be used to skip compacting the block, default is 90%
- `--verbose` GCV2 - enable more verbose output when running a GC pass
- Feature: Added new parameters for endpoint `admin/gc` (PUT)
- `compactblockthreshold` GCV2 - how much of a compact block should be used to skip compacting the block, default is 90%
- `verbose` GCV2 - enable more verbose output when running a GC pass
|
| | |
|
| |
|
|
|
| |
- Refactor GCV2 so GcReferencer::RemoveExpiredData returns a store compactor, moving out the actual disk work from deleting items in the index.
- Refactor GCV2 GcResult to reuse GcCompactStoreStats and GcStats
- Make Compacting of stores non-parallell to not eat all the disk I/O when running GC
|
| |
|
|
|
|
|
|
|
|
| |
initial version -- this is primarily intended to be used for running stress tests and/or benchmarks
example usage:
`zen run -n 10 -- zenserver-test`
`zen run -n 10 -- zenserver-test --ts=core.assert` run zenserver-test 10 times (testing only the `core.assert` test suite)
`zen run --time 600 --basepath=d:\test_dir\test1 -- zenserver-test` keeps spawning new instances for 10 minutes (600 seconds)
|
| |
|
| |
includes porting some compact binary builder code to use it since it had vestiges of the UE-side asserts
|
| | |
|
| |\ |
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| | |
* Added HTTP verb HEAD to objstroe get-object.
* Added HTTP verb HEAD to objstore get-object.
* Removed HEAD request exception for list-object.
|
| |/ |
|
| | |
|
| | |
|
| |
|
| |
* fix named event timout and test, fix blocking queue
|
| |
|
| |
this change replaces hard-coded port numbers in tests with dynamically assigned ports, to avoid potential issues around socket lifetimes and re-use policies
|
| | |
|
| |
|
|
|
|
|
|
|
| |
* make BlockingQueue::m_CompleteAdding non-atomic
* ZenCacheDiskLayer::Flush logging
* name worker threads in ZenCacheDiskLayer::DiscoverBuckets
* name worker threads in gcv2
* improved logging in ZenServerInstance
* scrub threadpool naming
* remove waitpid handling, we should just call wait to kill zombie processes
|
| |
|
| |
* if a file named root_manifest.ignore_schema_mismatch exists in the root data dir, ignore schema mismatch
|
| |
|
|
|
|
|
|
| |
* changed posix event implementation to use std::atomic instead of volatile
* ensure Event::Close() can take lock before deleting the inner object
* don't try to take the Event lock if the event is already signaled
* changed logic around Event::Wait without time-out. this works around some apparent issues on MacOS/Linux
* fix logic for posix process exit wait
|
| |
|
|
|
| |
this introduces a --snapshot-dir command line option to zenserver which specifies a directory which will be propagated to the persistence root directory on start-up.
This is most powerful with file systems which support block cloning, such as ReFS on Windows. This allows even very large state snapshots to be used repeatedly without having to worry about mutating the original dataset on disk. When using ReFS the state copy for even large state directories can be very fast since the duration is primarily proportional to the number of files in the tree rather than the size of the files being cloned. The storage requirements are also minimal as all data will be handled in a copy-on-write manner.
|
| | |
|