aboutsummaryrefslogtreecommitdiffstats
path: root/vespalib
Commit message (Collapse)AuthorAgeFilesLines
* Rename executorCount -> threadCountHenning Baldersheim2021-10-223-8/+8
|
* Add a metric for how many times a worker in a thread pool has woken up.Henning Baldersheim2021-10-224-24/+83
| | | | Also track the idle time a worker has and add metric for the utilization.
* expose fallback host if you ask nicelyHåvard Pettersen2021-10-214-9/+19
|
* add idle tracking helper classesHåvard Pettersen2021-10-216-0/+175
|
* Use a buffer on the heap over a buffer on the stack. Also increase buffer ↵Henning Baldersheim2021-10-201-4/+3
| | | | from 8k to 64k
* Use the ExecutorStats type directly.Henning Baldersheim2021-10-195-20/+17
|
* Move MonitoredRefCount and RetainGuard to vespalib.Tor Egge2021-10-184-0/+119
|
* add PrintTo for vespalib::stringArne H Juul2021-10-131-0/+9
|
* report more issuesHåvard Pettersen2021-10-131-1/+1
| | | | also drop the prefix of issues that are logged by the default handler
* Merge pull request #19514 from ↵Håvard Pettersen2021-10-123-0/+50
|\ | | | | | | | | vespa-engine/havardpe/classify-some-exceptions-as-unsafe central classification of 'unsafe' exceptions
| * central classification of 'unsafe' exceptionsHåvard Pettersen2021-10-123-0/+50
| |
* | Merge pull request #19510 from vespa-engine/arnej/add-binary-data-inputArne H Juul2021-10-122-20/+90
|\ \ | |/ |/| add "data" format as JSON extension
| * add "data" format as JSON extensionArne H Juul2021-10-122-20/+90
| | | | | | | | | | | | | | | | * when parsing JSON, allow binary data to be input as a hexdump prefixed with the letter "x"; this means that { foo: xFF0011 } will decode to an object where the "foo" field is of type DATA and contains the bytes 255, 0, and 17.
* | add more convenient issue reportingHåvard Pettersen2021-10-113-5/+55
|/
* Merge pull request #19457 from vespa-engine/havardpe/capture-issuesHåvard Pettersen2021-10-076-0/+210
|\ | | | | capture issues by binding handlers to threads
| * capture issues by binding handlers to threadsHåvard Pettersen2021-10-076-0/+210
| |
* | Update Verizon Media copyright notices.gjoranv2021-10-0792-92/+92
| |
* | Update 2019 Yahoo Holdings copyright notices.gjoranv2021-10-072-2/+2
| |
* | Update 2018 copyright notices.gjoranv2021-10-07115-115/+115
| |
* | Update 2017 copyright notices.gjoranv2021-10-07673-683/+683
|/
* run tests with various sizesArne H Juul2021-09-291-21/+38
| | | | * also use free() to keep valgrind happy
* file should reflect name of functionArne H Juul2021-09-294-3/+3
|
* add common binary_hamming_distance functionArne H Juul2021-09-286-0/+138
|
* log peer address when debug loggingArne Juul2021-08-261-1/+2
|
* Avoid narrowing.Tor Egge2021-08-231-1/+1
|
* Remove unused variables.Tor Egge2021-08-231-0/+5
|
* Extend test of array store compaction context.Tor Egge2021-08-181-10/+34
|
* Compact HNSW index when ratio of dead bytes / address space is too highTor Egge2021-08-182-0/+15
| | | | relative to used bytes / address space.
* Merge pull request #18752 from ↵Henning Baldersheim2021-08-161-0/+1
|\ | | | | | | | | vespa-engine/toregge/use-4096-buffers-for-hnsw-index-link-array-store Use 4096 buffers for HNSW link array store.
| * Use 4096 buffers for HNSW link array store.Tor Egge2021-08-161-0/+1
| | | | | | | | | | Configure link array store to handle arrays of 193 elements or less without indirect storage.
* | Instead of having one large array of individually allocated vectors useHenning Baldersheim2021-08-161-1/+0
|/ | | | | 2 large, optionally mmapped, vectors where the first just points into the second. In order to avoid resizing, count first.
* Assert that compacting flag is not already set for buffer thatTor Egge2021-08-161-4/+3
| | | | | is candiate for compaction. Use markCompacting() to mark buffer being compacted.
* Assert that compacting flag has been set on buffers when finishing compacting.Tor Egge2021-08-161-1/+5
| | | | Bypass buffer resize if it is already large enough.
* Consider reusing active buffer.Tor Egge2021-08-169-37/+138
|
* Revert "Consider reusing active buffer."Henning Baldersheim2021-08-159-138/+37
|
* Consider reusing active buffer.Tor Egge2021-08-139-37/+138
|
* Use stats from all active buffers of same type when resizing a buffer.Tor Egge2021-08-123-21/+21
|
* Use grow factor when considering switching to a new buffer.Tor Egge2021-08-114-12/+27
| | | | This enables earlier ramp-up of buffer sizes.
* Allow limits up to 512M.Henning Baldersheim2021-08-021-2/+3
|
* Generalize how address space usage is reported from attribute vectors.Geir Storli2021-07-152-0/+8
| | | | | This prepares for reporting address space usage for more attribute vector components, not only enum store and multi-value mapping.
* Merge pull request #18495 from vespa-engine/balder/allow-presizing-of-arraysHenning Baldersheim2021-07-0116-71/+98
|\ | | | | Balder/allow presizing of arrays
| * Move reserve method out the of the Value interface implementation.Henning Baldersheim2021-07-011-1/+1
| |
| * typedef -> usingHenning Baldersheim2021-07-011-1/+1
| |
| * Add final and noexceptHenning Baldersheim2021-07-018-37/+35
| |
| * Allow presizing of arrays in slime.Henning Baldersheim2021-07-0112-34/+63
| |
* | Undo unintended reformattingHenning Baldersheim2021-07-011-8/+12
| |
* | Inline write_cmpr_ulongHenning Baldersheim2021-07-011-34/+41
|/
* GC bad and unused method.Henning Baldersheim2021-07-012-4/+3
|
* Reduce exposure of internal details to reduce number of includes.Henning Baldersheim2021-06-3011-17/+21
|
* Use '= default'Henning Baldersheim2021-06-301-1/+1
|