summaryrefslogtreecommitdiffstats
path: root/searchcore
Commit message (Collapse)AuthorAgeFilesLines
* Fix forward declaration.Tor Egge2021-02-251-1/+1
|
* Avoid including singlenumericattribute.hppHenning Baldersheim2021-02-255-6/+0
|
* ensure we we abort operation if timestamp has changed between start and ↵Henning Baldersheim2021-02-243-8/+12
| | | | prepare. Also control the lifetime of the keys so they are either destructed or carried over to the next phase (completeMove) to ensure no races for the accounting.
* Also verify timestampHenning Baldersheim2021-02-241-1/+2
|
* Use current metadata when creating move operation.Henning Baldersheim2021-02-242-15/+36
|
* Obey stop signHenning Baldersheim2021-02-231-4/+7
|
* - Change error handling so that both synchonous and asynchronous errors can ↵Henning Baldersheim2021-02-236-43/+84
| | | | | | | | be reported back from bucket executor. - Treat remapping as an error. - For lidspace compaction job iterator is reset and will be recreated on next invocation. - For bucketmove th ebucket is rechecked and either discarded or restarted.
* Use a value metric for buckets_pending.Henning Baldersheim2021-02-232-2/+2
|
* Add metirc for buckets pending move.Henning Baldersheim2021-02-2314-22/+82
|
* move getting tensor into the attribute codeArne Juul2021-02-221-20/+3
| | | | | * avoids taking an extra copy of the tensor in the DirectTensorAttribute case
* Include memory and typeinfo when needed.Tor Egge2021-02-221-0/+1
|
* string.h does not need alloc.hHenning Baldersheim2021-02-2114-8/+20
|
* Merge pull request #16608 from ↵Henning Baldersheim2021-02-211-1/+1
|\ | | | | | | | | vespa-engine/toregge/explicitly-delete-implicitly-deleted-bucket-mover-move-constructor Explicitly delete implicitly deleted BucketMover move constructor.
| * Explicitly delete implicitly deleted BucketMover move constructor.Tor Egge2021-02-211-1/+1
| |
* | Include thread as needed.Tor Egge2021-02-201-0/+1
|/
* Only count the operation once.Henning Baldersheim2021-02-193-10/+7
| | | | Update test after code is fixed.
* Use a trinary to also handle temporary mismatching bucket used bits.Henning Baldersheim2021-02-1919-79/+96
|
* Merge pull request #16584 from vespa-engine/balder/some-code-healthHenning Baldersheim2021-02-192-23/+13
|\ | | | | - Reduce visibility of ClusterState and Distribution.
| * - Reduce visibility of ClusterState and Distribution.Henning Baldersheim2021-02-192-23/+13
| |
* | Only recompute bucketset to move when you are up.Henning Baldersheim2021-02-191-1/+1
|/
* Use braces and erase with iterator.Henning Baldersheim2021-02-181-4/+7
|
* Add alternative bucket move job without frozen concept.Henning Baldersheim2021-02-1712-100/+1239
|
* Merge pull request #16495 from vespa-engine/arnej/add-size-literalsArne H Juul2021-02-1634-95/+113
|\ | | | | add literals for sizes in KB, MB, and GB
| * use 4_Gi directlyArne Juul2021-02-151-2/+1
| |
| * use 15_Gi/17_Gi directlyArne Juul2021-02-151-4/+3
| |
| * use size literals in searchcoreArne Juul2021-02-1534-91/+111
| |
* | Setup MmapFileAllocatorFactory.Tor Egge2021-02-151-0/+2
|/
* When making callbacks like notifying about new bucket show the gurad that is ↵Henning Baldersheim2021-02-1457-283/+251
| | | | | | held. Then this guard can used instead of possibly making a deadlock if trying to take it yourself.
* Add alloc_aligned member function.Tor Egge2021-02-131-2/+1
|
* Forward declare MemoryAllocator in vespalib/util/alloc.h.Tor Egge2021-02-131-0/+1
|
* Merge pull request #16501 from vespa-engine/balder/less-assert-in-header-filesHenning Baldersheim2021-02-1311-47/+26
|\ | | | | Less asserts in header files.
| * Less asserts in header files.Henning Baldersheim2021-02-1311-47/+26
| |
* | urgen ->urgentHenning Baldersheim2021-02-131-1/+1
| |
* | Move control of internal limits closer to where the limitation is.Henning Baldersheim2021-02-133-14/+10
|/
* Add unit test for cappingHenning Baldersheim2021-02-111-0/+24
|
* Ensure that we do not go above 16G as we then might trigger other internal ↵Henning Baldersheim2021-02-111-6/+7
| | | | | | limits that we do not have full control of.
* Merge pull request #16469 from vespa-engine/balder/refactor-moverHenning Baldersheim2021-02-119-126/+194
|\ | | | | Refactor to splitt out work that can be done in separate thread.
| * Refactor to splitt out work that can be done in separate thread.Henning Baldersheim2021-02-109-126/+194
| |
* | Wire in configuration for bucket move jobHenning Baldersheim2021-02-109-102/+144
|/
* To ensure that lid stats is sampled after everything has been completed ↵Henning Baldersheim2021-02-072-0/+2
| | | | | | compaction must happen in 2 stages. First it must reach inSync, then it must rescheduled once more in the master thread. Then all movement is visible in the master thread.
* Use gtestHenning Baldersheim2021-02-047-634/+640
|
* Call std::_Exit(0) if SIGINT or SIGTERM is received while loadingTor Egge2021-02-041-1/+43
| | | | data structures, replaying transaction log or performing startup reprocessing.
* Split to one test file per fixture.Henning Baldersheim2021-02-043-132/+151
|
* Refactor test for easier reuse.Henning Baldersheim2021-02-045-508/+565
|
* Merge pull request #16367 from vespa-engine/toregge/include-thread-as-neededHenning Baldersheim2021-02-033-0/+3
|\ | | | | Include thread as needed.
| * Include thread as needed.Tor Egge2021-02-033-0/+3
| |
* | Count on the right places.Henning Baldersheim2021-02-031-2/+13
| |
* | Remove the sync interface from Bucket Executor. Due to the nature of ↵Henning Baldersheim2021-02-027-32/+36
|/ | | | | | requiring a bucket lock it is very hard to get sync to work in a safe way. Instead the users must do their own accounting as they know their own threading model.
* Merge pull request #16348 from vespa-engine/balder/refactor-bucketmove-jobHenning Baldersheim2021-02-027-94/+98
|\ | | | | Factor out the scan iterator from the BucketMoveJob.
| * Factor out the scan iterator from the BucketMoveJob.Henning Baldersheim2021-02-027-94/+98
| |