Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Remove unused wrapper type for disk count. | Tor Egge | 2020-11-06 | 1 | -1/+0 |
| | |||||
* | Remove disk config from ServiceLayerNode and StorageNode. | Tor Egge | 2020-11-06 | 2 | -23/+2 |
| | |||||
* | Remove disk distribution from C++ part of vdslib. | Tor Egge | 2020-11-05 | 1 | -9/+1 |
| | |||||
* | Merge pull request #15144 from ↵ | Tor Brede Vekterli | 2020-11-04 | 6 | -11/+112 |
|\ | | | | | | | | | vespa-engine/vekterli/increase-priority-for-global-bucket-merges Increase priority for global bucket merges | ||||
| * | Make prioritization of global bucket merges live configurable | Tor Brede Vekterli | 2020-11-03 | 6 | -3/+63 |
| | | | | | | | | | | | | Lets an application that is starving default space merges explicitly disable prioritization temporarily. This is a stop-gap before we add more robust and generalized handling of resource exhaustion scenarios. | ||||
| * | Increase priority for global bucket merges | Tor Brede Vekterli | 2020-11-02 | 6 | -11/+52 |
| | | | | | | | | | | | | | | | | | | To avoid global buckets competing with (and usually being starved by) default bucket space merges, explicitly prioritize default bucket merges above most other load. Increases both distributor-internal maintenance schedulinr priority and persistence-level operation priority. | ||||
* | | Remove test that mentions an explicit disk state. | Tor Egge | 2020-11-03 | 1 | -4/+0 |
| | | |||||
* | | Update log message to not mention partitions. | Tor Egge | 2020-11-03 | 1 | -1/+1 |
| | | |||||
* | | Remove diskstate from vdslib. | Tor Egge | 2020-11-03 | 9 | -196/+7 |
| | | |||||
* | | Track parameter name in member function definition. | Tor Egge | 2020-11-03 | 1 | -1/+1 |
| | | |||||
* | | Remove stray template keyword. | Tor Egge | 2020-11-02 | 1 | -1/+1 |
| | | |||||
* | | Merge pull request #15137 from ↵ | Tor Brede Vekterli | 2020-11-02 | 10 | -31/+242 |
|\ \ | |/ |/| | | | | | vespa-engine/toregge/apply-diff-entry-using-async-spi-methods-during-bucket-merge Apply diff entries using async spi methods during bucket merge. | ||||
| * | Move wait out of check_result. | Tor Egge | 2020-11-02 | 6 | -24/+39 |
| | | | | | | | | | | Pass elapsed time along with result and update metric after having checked the result. | ||||
| * | Apply diff entries using async spi methods during bucket merge. | Tor Egge | 2020-11-02 | 10 | -31/+227 |
| | | |||||
* | | Update and add comments | Tor Brede Vekterli | 2020-11-02 | 3 | -8/+19 |
| | | |||||
* | | Add test for explicit read guard iterator key ordering | Tor Brede Vekterli | 2020-10-30 | 1 | -0/+23 |
| | | |||||
* | | Add stripe bits config and wire to implementation | Tor Brede Vekterli | 2020-10-30 | 16 | -25/+59 |
| | | | | | | | | | | Default is zero bits, which causes the standard, non-striped implementation to be used. | ||||
* | | Add striped implementation of B-tree content node bucket database | Tor Brede Vekterli | 2020-10-30 | 17 | -76/+517 |
|/ | | | | | | | | | | Abstracts away multiple underlying B-tree DBs that each hold a subset of the super bucket space. Offers ordered iteration via a priority-queue based view over the sub DBs. Not yet ready for prime time, as the striping inherently requires an absolute lower bound on the bucket bits used in the system, which is currently not enforced. | ||||
* | Use source-only nodes again during bucket merge if hasMask histogram shows | Tor Egge | 2020-10-29 | 4 | -54/+99 |
| | | | | that some diff entries are unavailable on all other nodes. | ||||
* | Merge pull request #15074 from vespa-engine/toregge/keep-unmapped-hasmask-bits | Henning Baldersheim | 2020-10-29 | 5 | -4/+35 |
|\ | | | | | Keep hasMask bits for nodes not involved in merge operation step | ||||
| * | Keep hasMask bits for nodes not involved in merge operation step. | Tor Egge | 2020-10-28 | 5 | -4/+35 |
| | | |||||
* | | GC unused content bucket btree flag | Henning Baldersheim | 2020-10-28 | 1 | -1/+1 |
|/ | |||||
* | Rename MaskRemapper to HasMaskRemapper. | Tor Egge | 2020-10-28 | 6 | -60/+116 |
| | | | | | Move HasMaskRemapper to separate file. Add separate unit test for HasMaskRemapper. | ||||
* | Remove dead code for unused tool | Tor Brede Vekterli | 2020-10-28 | 1 | -518/+0 |
| | |||||
* | Remove legacy Judy array-backed bucket DB implementation | Tor Brede Vekterli | 2020-10-28 | 17 | -2233/+7 |
| | |||||
* | Merge pull request #15041 from ↵ | Tor Brede Vekterli | 2020-10-27 | 4 | -11/+268 |
|\ | | | | | | | | | vespa-engine/toregge/add-unit-test-for-bucket-merge-with-partially-filled-diff-from-last-source-only-node Add unit test for bucket merge with partially filled diff | ||||
| * | Add comments about per-reply masks. | Tor Egge | 2020-10-27 | 3 | -1/+16 |
| | | |||||
| * | Add unit test for bucket merge with partially filled diff from last source ↵ | Tor Egge | 2020-10-26 | 4 | -11/+253 |
| | | | | | | | | only node. | ||||
* | | - Honour max merge chunk size in one place. | Henning Baldersheim | 2020-10-26 | 6 | -51/+19 |
|/ | | | | - GC incomplete enable_merge_local_node_choose_docs_optimalization handling. | ||||
* | Merge pull request #15033 from ↵ | Geir Storli | 2020-10-26 | 8 | -23/+245 |
|\ | | | | | | | | | vespa-engine/geirst/async-message-handling-on-schedule Add support for async message handling when scheduling storage messag… | ||||
| * | Use noexcept and prefix iterator increment. | Geir Storli | 2020-10-26 | 3 | -3/+3 |
| | | |||||
| * | Add support for async message handling when scheduling storage messages in ↵ | Geir Storli | 2020-10-26 | 8 | -23/+245 |
| | | | | | | | | | | | | | | | | | | FileStorManager. When turned on, the calling thread (e.g. FNET network thread when using Storage API RPC) gets the next async message to handle (if any) as part of scheduling a storage message. This async message is then handled by the calling thread immediately, instead of going via a persistence thread. | ||||
* | | Merge pull request #15032 from ↵ | Henning Baldersheim | 2020-10-26 | 1 | -1/+1 |
|\ \ | | | | | | | | | | | | | vespa-engine/vekterli/avoid-mergehandler-spam-log-level-crash Don't try to access container element prior to adding it when spam logging | ||||
| * | | Don't try to access container element prior to adding it when spam logging | Tor Brede Vekterli | 2020-10-26 | 1 | -1/+1 |
| |/ | |||||
* / | Drop immediate commit, as it is gone. | Henning Baldersheim | 2020-10-23 | 1 | -1/+1 |
|/ | |||||
* | Use default destructor outside class definition. | Geir Storli | 2020-10-22 | 1 | -2/+6 |
| | |||||
* | Reduce code duplication by consolidating setup of PersistenceHandler. | Geir Storli | 2020-10-22 | 1 | -39/+38 |
| | |||||
* | Reduce code duplication by consolidating setup of FileStorHandler. | Geir Storli | 2020-10-22 | 1 | -160/+76 |
| | |||||
* | Reduce code duplication by using TestFileStorComponents in more tests. | Geir Storli | 2020-10-22 | 1 | -121/+39 |
| | |||||
* | Test that new repo is reflected in cached value in PersistenceUtil. | Henning Baldersheim | 2020-10-21 | 1 | -1/+15 |
| | |||||
* | Add generation counter to StorageComponent to detect if repos have changed ↵ | Henning Baldersheim | 2020-10-20 | 9 | -25/+52 |
| | | | | | | and need reload. Happy path where you do not need any reload is a single atomic relaxed load as opposed to at least 3 atomic locked instructions. | ||||
* | Merge pull request #14970 from vespa-engine/balder/gc-unused-disk-count-code | Henning Baldersheim | 2020-10-20 | 14 | -107/+78 |
|\ | | | | | GC unused disk dimension. | ||||
| * | Use simple assert instead | Henning Baldersheim | 2020-10-20 | 1 | -3/+1 |
| | | |||||
| * | GC unused disk dimension. | Henning Baldersheim | 2020-10-20 | 14 | -106/+79 |
| | | |||||
* | | Fix forward declarations. | Tor Egge | 2020-10-20 | 6 | -6/+6 |
| | | |||||
* | | Merge pull request #14979 from vespa-engine/toregge/remove-spi-partitionid-type | Henning Baldersheim | 2020-10-20 | 3 | -12/+5 |
|\ \ | | | | | | | Remove storage::spi::PartitionId type. | ||||
| * | | Remove storage::spi::PartitionId type. | Tor Egge | 2020-10-20 | 3 | -12/+5 |
| |/ | |||||
* / | Allow for faster access of thread local variable. | Henning Baldersheim | 2020-10-20 | 2 | -2/+8 |
|/ | | | | | This is possible since we do not load our code explicit as dynamic library. Also replace __thread by standard c++ linkage thread_local. | ||||
* | Do not access an object that has been moved away. | Henning Baldersheim | 2020-10-20 | 1 | -1/+1 |
| | |||||
* | Take number of response threads and number of network threads into account. | Henning Baldersheim | 2020-10-20 | 2 | -1/+4 |
| |