summaryrefslogtreecommitdiffstats
path: root/slobrok
Commit message (Collapse)AuthorAgeFilesLines
* Include mutex to get declaration of std::mutexTor Egge2022-09-251-0/+1
|
* Reduce include and visibility of ptrholder.hHenning Baldersheim2022-09-231-2/+1
|
* remove unused doxygen setup filesArne Juul2022-08-291-228/+0
|
* Add capability request filters to Slobrok RPCsTor Brede Vekterli2022-08-262-0/+28
|
* Deinline ServiceMapping copy constructor and copy assignment.Tor Egge2022-05-302-0/+6
|
* Fold staging_vespalib into vespalibHenning Baldersheim2022-05-201-1/+0
|
* make _reqDone atomicHåvard Pettersen2022-05-194-10/+11
|
* Move state_server, metrivs and some all executors from staging_vespalib too ↵Henning Baldersheim2022-05-183-5/+5
| | | | vespalib.
* wait for more things before destructing transportHåvard Pettersen2022-05-022-0/+2
| | | | | | Exiting the main loop of the transport thread(s) does no longer mean you are completely finished. We also need to wait for async dns resolving and tls handshake work.
* Initialize transitive ctor dependencies before useTor Brede Vekterli2022-04-272-7/+8
| | | | | Ensure that SBEnv fields that are accessed by RPCHooks ctor are initialized before the RPCHooks field itself is created.
* remove FastOS_ApplicationHåvard Pettersen2022-04-085-79/+76
| | | | | | | | fixup (per application): - maybe ignore SIGPIPE - wire argc/argv untangle Vespa Test Framework strip down deprecated TestApp
* use getopt/getopt_long directlyHåvard Pettersen2022-04-012-11/+9
|
* Revert "Fix Slobrok race between completion callback and scheduled task ↵Tor Brede Vekterli2022-03-041-37/+37
| | | | [run-systemtest]"
* Minor code cleanups - no change in semanticsTor Brede Vekterli2022-02-281-33/+30
|
* Fix Slobrok race between completion callback and scheduled taskTor Brede Vekterli2022-02-281-4/+7
| | | | | | This resolves a racing read/write on "request done" status that may happen when RPC `HandleDone()` is invoked from one FNET thread and `PerformTask()` is concurrently invoked from another FNET thread.
* Add noexcept specifiers (slobrok).Tor Egge2022-02-262-2/+2
|
* Revert "Revert "- Create the common transport and threadpool in the main loop.""Henning Baldersheim2022-02-242-4/+7
|
* Revert "- Create the common transport and threadpool in the main loop."Henning Baldersheim2022-02-242-7/+4
|
* - Create the common transport and threadpool in the main loop.Henning Baldersheim2022-02-242-4/+7
| | | | | - Also use the common transport for config subscriptions. - Put The TransportConfig in the fnet namespace.
* Make ConfigUri constructors explicit and use same context where possible in ↵Henning Baldersheim2022-02-201-1/+1
| | | | proton.
* Reduce code visibility and include only what you need from config library.Henning Baldersheim2022-02-067-16/+29
|
* _executor -> _threadHenning Baldersheim2021-12-091-6/+3
|
* Add init_fun to vespalib::Thread too to figure out what the thread is used for.Henning Baldersheim2021-12-091-2/+4
|
* Update Verizon Media copyright notices.gjoranv2021-10-0736-36/+36
|
* Update 2017 copyright notices.gjoranv2021-10-0773-73/+73
|
* Fix format strings.Tor Egge2021-09-271-2/+2
|
* remove wiring of useNewLogic flagArne H Juul2021-09-205-34/+12
|
* simplify moreArne H Juul2021-09-202-10/+6
|
* remove unused filesArne H Juul2021-09-2011-705/+0
|
* match in unnamed namespaceArne H Juul2021-09-202-39/+19
|
* removing more codeArne H Juul2021-09-2012-144/+35
|
* remove unused variablesArne H Juul2021-09-207-28/+8
|
* stop sending old-sync RPC callsArne H Juul2021-09-203-193/+4
|
* remove unused variablesArne H Juul2021-09-203-6/+4
|
* no double logic if feature flag is onArne H Juul2021-09-202-5/+7
|
* test rpc_mapping_monitorHåvard Pettersen2021-09-204-1/+235
| | | | using the new fnet::TransportDebugger tool
* Merge pull request #19182 from vespa-engine/arnej/move-some-codeArne H Juul2021-09-2011-84/+44
|\ | | | | Arnej/move some code
| * use _exchanger.removePartner for symmetryArne H Juul2021-09-171-6/+2
| |
| * remove unused RpcServerMap::lookupPattern, move match to RPCHooksArne H Juul2021-09-174-38/+21
| |
| * more peer management logic directly in SBEnvArne H Juul2021-09-177-41/+22
| |
* | Merge pull request #19192 from vespa-engine/arnej/tune-timingArne H Juul2021-09-171-2/+2
|\ \ | | | | | | use same timing for initial health check
| * | use same timing for initial health checkArne H Juul2021-09-171-2/+2
| |/ | | | | | | | | | | * use the 2.7s +- 0.2s timing for first health check, that is also used for subsequent health checks * also, set timeout for ping to the usual 5 seconds value
* / no need to check useNewLogic() hereArne H Juul2021-09-171-10/+1
|/
* Merge pull request #19171 from vespa-engine/arnej/always-use-new-logic-2Håvard Pettersen2021-09-161-181/+17
|\ | | | | cosmetic code style changes
| * cosmetic code style changesArne H Juul2021-09-161-181/+17
| |
* | low-level testing of fnet enabled byHåvard Pettersen2021-09-161-1/+1
|/ | | | | | | | | | | | | | | | | | | 1. Manipulating how much time is spent waiting for events while also manipulating how the passing of time is observed by the transport threads. (fnet::TimeTools) 2. Periodic call-backs from FNET_Transport. All transport threads are blocked during these call-backs. Between each call-back, all transport threads perform a single event loop iteration and any async operations not performed in transport threads are flushed. (FNET_Transport::attach_capture_hook) 3. Combining 1 and 2 in a way that lets the test code run all transport threads one iteration at a time, letting the transport threads run concurrently with each other, but no transport thread is allowed to run concurrently with the test code. (fnet::TransportDebugger)
* always run with new logicArne H Juul2021-09-162-5/+5
| | | | | * we decided to avoid any warnings on unregister, so update unit test accordingly.
* simplify request completion handlingArne H Juul2021-09-157-31/+77
|
* avoid spurious errorArne H Juul2021-09-101-8/+2
| | | | | | * last changes triggered this warning: slobrok.unregisterRpcServer(foo -> tcp/bar:4294967295) failed: conflict detected * why it wants to unregister "-1" as port number is unclear, but it's a no-op anyway
* take no local action in new doAdd()Arne H Juul2021-09-101-2/+0
|