summaryrefslogtreecommitdiffstats
path: root/vespalib
Commit message (Collapse)AuthorAgeFilesLines
* Merge pull request #21209 from ↵Håvard Pettersen2022-02-161-4/+13
|\ | | | | | | | | vespa-engine/vekterli/well-defined-race-cond-checking-in-spinlock-test Use atomic refs to make spin lock race condition testing well-defined
| * Use atomic refs to make spin lock race condition testing well-definedTor Brede Vekterli2022-02-151-4/+13
| | | | | | | | Relaxed loads/stores give the same effect but without the undefined-ness.
* | Merge pull request #21207 from vespa-engine/havardpe/better-error-messagesLester Solbakken2022-02-162-4/+29
|\ \ | |/ |/| better error messages
| * better error messagesHåvard Pettersen2022-02-152-4/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | added support for capturing stderr when you expect stuff to fail in order to capture all error messages from the child process. simplify ServerCmd to always be verbose (print stdin/stdout(stderr) interactions). improve ServerCmd to enable explicitly checking the child process exit code. fixed test dependency on binary for eval expr and analyze model tests.
* | Disable use of nice() on darwin.Tor Egge2022-02-152-1/+5
|/
* Include needed header files.Tor Egge2022-02-142-0/+2
|
* Merge pull request #21147 from ↵Geir Storli2022-02-112-3/+3
|\ | | | | | | | | vespa-engine/toregge/move-memory-allocator-to-be-last-argument-for-unique-store-constructor Move memory allocator argument to last position in unique store constructor.
| * Move memory allocator argument to last position in unique store constructor.Tor Egge2022-02-112-3/+3
| |
* | Merge pull request #21136 from vespa-engine/havardpe/be-niceHåvard Pettersen2022-02-114-36/+135
|\ \ | | | | | | added support for adjusting the nice value of threads
| * | added support for adjusting the nice value of threadsHåvard Pettersen2022-02-104-36/+135
| | | | | | | | | | | | ... using a thread init function wrapper
* | | Move memory allocator argument to be last argument.Tor Egge2022-02-113-10/+9
| | |
* | | Add memory allocator to array store.Tor Egge2022-02-1111-50/+206
| |/ |/|
* | Add missing override.Tor Egge2022-02-101-1/+1
| |
* | Add memory allocator to unique store allocatorTor Egge2022-02-1016-33/+160
|/
* Keep using same memory allocator when resizing rcu vector.Tor Egge2022-02-093-3/+76
|
* Add reset and create methods to vespalib::Array.Tor Egge2022-02-094-0/+72
|
* Block in-place resize from nonzero to zero size.Tor Egge2022-02-092-19/+42
|
* Merge pull request #21115 from vespa-engine/balder/use-mmap-for-large-vectors-5Håvard Pettersen2022-02-091-6/+8
|\ | | | | Automatically switch to mmap for large allocations.
| * Automatically switch to mmap for large allocations.Henning Baldersheim2022-02-091-6/+8
| |
* | Merge pull request #21112 from vespa-engine/balder/use-mmap-for-large-vectors-3Håvard Pettersen2022-02-093-6/+9
|\ \ | | | | | | vector of string_id tends to become very large. Use mmap allocation a…
| * | vector of string_id tends to become very large. Use mmap allocation ↵Henning Baldersheim2022-02-083-6/+9
| | | | | | | | | | | | automatically.
* | | Merge pull request #21111 from vespa-engine/balder/use-mmap-for-large-vectors-2Håvard Pettersen2022-02-091-1/+2
|\ \ \ | |_|/ |/| | Use mmap automatically for large vectors.
| * | Use mmap automatically for large vectors.Henning Baldersheim2022-02-081-1/+2
| |/
* / - Generate mode modern c++ code.Henning Baldersheim2022-02-083-32/+6
|/ | | | | - Use existing predefined types. - Allow for better management by allowing large string vectors to be mmapped.
* Reduce code visibility and include only what you need from config library.Henning Baldersheim2022-02-061-63/+104
|
* Merge pull request #21076 from vespa-engine/balder/use-auto-allocatorHenning Baldersheim2022-02-041-10/+14
|\ | | | | Automatic select between heap or mmap allocation
| * Automatic select between heap or mmap allocationHenning Baldersheim2022-02-041-10/+14
| |
* | test per-thread nice valuesHåvard Pettersen2022-02-043-0/+53
|/
* Tag all document store tasks executed on the shared executor with cpu category.Geir Storli2022-01-311-0/+2
|
* Include <array> when using std::array.Tor Egge2022-01-281-0/+1
|
* Merge pull request #20958 from vespa-engine/havardpe/ignore-coresHåvard Pettersen2022-01-281-0/+1
|\ | | | | ignore local core files produced by gtest during death testing
| * ignore local core files produced by gtest during death testingHåvard Pettersen2022-01-271-0/+1
| |
* | Tag proton thread pools with CPU categories.Geir Storli2022-01-271-0/+1
|/
* Merge pull request #20937 from vespa-engine/havardpe/tag-cpu-usageGeir Storli2022-01-266-17/+59
|\ | | | | tag match/summary engines with READ cpu usage
| * tag match/summary engines with READ cpu usageHåvard Pettersen2022-01-266-17/+59
| | | | | | | | tag llvm compilation tasks with SETUP cpu usage
* | Support live reconfig of dynamic persistence throttlingTor Brede Vekterli2022-01-262-13/+60
|/ | | | | | | | Start moving internal `stor-server` throttling config to use struct config instead of separate fields, as this is more flexible and better matches how we configure throttling elsewhere. For now, let dynamic throttling be enabled via both the new and the old config enum. Config model will be updated to use the new config struct shortly.
* Merge pull request #20915 from vespa-engine/havardpe/proton-cpu-util-metricsGeir Storli2022-01-253-63/+68
|\ | | | | added cpu util metrics
| * added cpu util metricsHåvard Pettersen2022-01-243-63/+68
| |
* | Don't allow pending ops in DynamicOperationThrottler when destroying instance.Tor Egge2022-01-241-1/+4
|/
* use pthread clock for total cpu usage as wellHåvard Pettersen2022-01-243-26/+16
|
* Merge pull request #20903 from vespa-engine/havardpe/improve-cpu-usage-trackingHåvard Pettersen2022-01-246-138/+446
|\ | | | | improve cpu usage tracking
| * explicitly wait for conflicts to happenHåvard Pettersen2022-01-213-3/+28
| |
| * improve cpu usage trackingHåvard Pettersen2022-01-216-137/+420
| | | | | | | | | | | | | | | | | | | | | | - add cpu_usage::RUsage primitive used for getrusage - reduce the number of categories (small initial set) - use RUsage to fill OTHER category (more complete samples) - threads no longer track OTHER category explicitly (less overhead) - changing cpu category to itself is now nop (less overhead) - make ThreadTracker and SampleConflict private (less API clutter) - improved testing (sample conflict resolution) - improves testing (thread tracker implementation)
* | Replace storage operation throttler with vespalib implementationTor Brede Vekterli2022-01-211-8/+115
|/ | | | Also move the remaining throttler unit tests to vespalib.
* Merge pull request #20888 from ↵Tor Brede Vekterli2022-01-216-0/+587
|\ | | | | | | | | vespa-engine/vekterli/add-operation-throttler-to-vespalib Add SharedOperationThrottler to vespalib utils
| * Add reference to Java throttle policy fileTor Brede Vekterli2022-01-211-1/+2
| |
| * Add SharedOperationThrottler to vespalib utilsTor Brede Vekterli2022-01-206-0/+586
| | | | | | | | | | | | A cleaned up, de-MessageBus'd version of the implementation that already exists in the storage module. The old implementation will be replaced with the vespalib one very soon.
* | benchmark std::atomic wait/notify wakeupHåvard Pettersen2022-01-201-4/+38
|/ | | | also measure cpu usage for waiting threads
* Merge pull request #20815 from ↵Henning Baldersheim2022-01-173-0/+511
|\ | | | | | | | | vespa-engine/havardpe/track-multi-threaded-cpu-usage-per-category track cpu usage across threads and usage categories
| * correlate sampled cpu usage with rusageHåvard Pettersen2022-01-171-6/+21
| |