aboutsummaryrefslogtreecommitdiffstats
path: root/searchcore/src
Commit message (Collapse)AuthorAgeFilesLines
* Remove broken copy and move constructor and assignment operatoos on ↵Henning Baldersheim2020-10-071-1/+2
| | | | | | vespalib::Lock and vespalib::Monitor. Also repair broken usages of the same.
* Add noexcept as indicated by -WnoeexceptHenning Baldersheim2020-10-0731-159/+109
|
* Take bucket guard after having passed validLid() check.Tor Egge2020-10-061-1/+1
|
* Remove unneeded include.Tor Egge2020-10-061-1/+0
|
* Reuse document meta store state from prepare step instead of doingTor Egge2020-10-0622-144/+170
| | | | a new lookup in btree mapping from gid to lid during live feed.
* Remove document meta store stub method.Tor Egge2020-10-052-62/+1
|
* Merge pull request #14714 from vespa-engine/geirst/feed-bm-rpc-targets-per-nodeGeir Storli2020-10-051-0/+16
|\ | | | | Add option to set RPC targets per node.
| * Add option to set RPC targets per node.Geir Storli2020-10-051-0/+16
| |
* | Explicitly instantiate common attribute update member functions.Tor Egge2020-10-055-5/+2
|/
* Merge pull request #14710 from ↵Henning Baldersheim2020-10-041-7/+4
|\ | | | | | | | | vespa-engine/balder/use-shared-executor-when-reading-from-summary-store Use shared executor when accessing docstore.
| * Use shared executor when accessing docstore.Henning Baldersheim2020-10-041-7/+4
| |
* | Merge pull request #14683 from ↵Geir Storli2020-10-041-3/+5
|\ \ | |/ |/| | | | | vespa-engine/geirst/support-multiple-rpc-targets-in-storage-api-over-rpc Add support for multiple RPC targets per node when using Storage API …
| * Add support for multiple RPC targets per node when using Storage API over RPC.Geir Storli2020-10-021-3/+5
| | | | | | | | | | | | | | This should allow for better parallelization and higher feed throughput. The bucket id associated with a message is used to select the RPC target. This ensures the same RPC target is used for all messages to the same bucket to the same node, and the RPC target itself handles sequencing of these messages.
* | Include vector when using std::vector.Tor Egge2020-10-041-0/+1
| |
* | Modify applyCommit to commit without updating synctoken if too early.Henning Baldersheim2020-10-033-4/+7
| |
* | Use correct serial number when creating attribute.Henning Baldersheim2020-10-037-17/+13
| |
* | Add description of max-pending parameter, and keep ordering.Henning Baldersheim2020-10-031-26/+27
| | | | | | | | Must have consistent ordering
* | Add maxpending which by default is 1000 to control max inflight.Henning Baldersheim2020-10-031-15/+24
| |
* | Add wakeup call on the executors to allow to urgent wakeup of lazy exeutors ↵Henning Baldersheim2020-10-037-0/+14
| | | | | | | | optimized for throughput.
* | Merge pull request #14688 from vespa-engine/balder/use-forground-executorHenning Baldersheim2020-10-021-9/+4
|\ \ | | | | | | Use foreground executor.
| * | Use foreground executor.Henning Baldersheim2020-10-021-9/+4
| | |
* | | Add option to control sequencer used for attributesHenning Baldersheim2020-10-021-7/+22
|/ /
* | Don't name unused arguments.Tor Egge2020-10-021-9/+3
| |
* | Use default destructor.Tor Egge2020-10-021-3/+1
| |
* | Enable feeding over message bus for vespa-feed-bm.Tor Egge2020-10-0217-83/+685
|/
* Update usage message. Style fix.Tor Egge2020-10-011-3/+5
|
* Fix forward declaration of BmStorageLinkContext.Tor Egge2020-10-011-1/+1
|
* Calculate average throughput and cleanup how benchmark results are printed.Geir Storli2020-10-011-19/+80
|
* Rename 'threads' to 'client-threads' and change default to 1.Geir Storli2020-10-011-21/+21
|
* Add option to set num response threads used in service layer.Geir Storli2020-10-011-1/+15
|
* Use btree bucket db as default.Geir Storli2020-10-011-11/+24
|
* Use the same RPC compression config as the default in ↵Geir Storli2020-10-013-55/+67
| | | | stor-communicationmanager.def.
* Remove commented out includes.Tor Egge2020-09-301-4/+0
|
* Move out classes to separate files. Enable feeding via distributor.Tor Egge2020-09-3016-179/+417
|
* Fix format string.Tor Egge2020-09-291-1/+1
|
* Avoid overlapping buckets between bm feed threads.Tor Egge2020-09-291-26/+47
|
* Adjust log context.Tor Egge2020-09-281-1/+1
|
* Check storage reply for errors in vespa-feed-bm.Tor Egge2020-09-2811-13/+110
|
* Add distributor "process" to vespa-feed-bm.Tor Egge2020-09-281-32/+158
|
* Drop commit of memory structures if high visibility-delayHenning Baldersheim2020-09-287-2/+21
|
* - Amortise write cost by grouping mulptiple operations together when writing ↵Henning Baldersheim2020-09-284-6/+50
| | | | | | | | | to TLS. - Commit memorystructures only then persisting to disk. - Ack operations back to user when both are completed. - Do not schedule a new commit task until both the tls and the memory structures have been comitted.
* Merge pull request #14572 from ↵Henning Baldersheim2020-09-284-12/+260
|\ | | | | | | | | vespa-engine/toregge/pass-storage-api-messages-directly-to-storage-chain Add option to feed using storage api protocol directly on storage chain.
| * Use context to keep track of extra storagelink instance used toTor Egge2020-09-283-18/+46
| | | | | | | | inject benchmark feed.
| * Add option to feed using storage api protocol directly onTor Egge2020-09-254-12/+232
| | | | | | | | storage chain.
* | Spam to debugHenning Baldersheim2020-09-272-2/+2
| |
* | Add logging of forcecommit calls.Henning Baldersheim2020-09-271-1/+1
| |
* | Ensure that contexts have the proper lifetime.Henning Baldersheim2020-09-263-2/+6
| |
* | Drop virtual keyword on override.Henning Baldersheim2020-09-261-3/+3
| |
* | Allow early ack on non-zero visibility delay.Henning Baldersheim2020-09-261-1/+1
| |
* | Separate immediateCommit and allowEarlyAckHenning Baldersheim2020-09-268-16/+24
| |