* functions: modify datastore to accomodate hot containers support
* functions: protocol between functions and hot containers
* functions: add hot containers clockwork
* fn: add hot containers support
* functions: add bounded concurrency
* functions: plug runners to sync and async interfaces
* functions: update documentation about the new env var
* functions: fix test flakiness
* functions: the runner is self-regulated, no need to set a number of runners
* functions: push the execution to the background on incoming requests
* functions: ensure async tasks are always on
* functions: add prioritization to tasks consumption
Ensure that Sync tasks are consumed before Async tasks. Also, fixes
termination races problems for free.
* functions: remove stale comments
* functions: improve mem availability calculation
* functions: parallel run for async tasks
* functions: check for memory availability before pulling async task
* functions: comment about rnr.hasAvailableMemory and sync.Cond
* functions: implement memory check for async runners using Cond vars
* functions: code grooming
- remove unnecessary goroutines
- fix stale docs
- reorganize import group
* Revert "functions: implement memory check for async runners using Cond vars"
This reverts commit 922e64032201a177c03ce6a46240925e3d35430d.
* Revert "functions: comment about rnr.hasAvailableMemory and sync.Cond"
This reverts commit 49ad7d52d341f12da9603b1a1df9d145871f0e0a.
* functions: set a minimum memory availability for sync
* functions: simplify the implementation by removing the priority queue
* functions: code grooming
- code deduplication
- review waitgroups Waits
* Always decrement memory even if task preparation or execution fails.
* Fall back to max 2GB memory on non-Linux. 300GB is ridiculous.
* Simplify loop
Currently, async workers are started before HTTP interface is available
to get their requests. It fixes by ensuring that async workers are
started after HTTP interface is up.
Essentially we are getting rid of an error message during bootstrap:
ERRO[0000] Could not fetch task error=Get http://127.0.0.1:8080/tasks: dial tcp 127.0.0.1:8080: getsockopt: connection refused
* Reduce test verbosity
* Divert gin's log to the test buffer
* Divert stdlib's log to the test buffer
* Add bolt tests into log buffer
* Add a linebreak to improve log output layout
* Update Readme and add Makefile
* Skip stale tests (in wait for stdin support)
* Revert "Skip stale tests (in wait for stdin support)"
This reverts commit 228da3776503f40ca53df70a79a9e4a9c73fd8b5.
By default, BoltDB will hang while waiting to acquire lock to the
datafile, thus the users might find themselves waiting for something
but not what. The added timeout aims inform use about what's
happening.
Also this renames MQADR to TASKSRV, refactor configuration to read
environment variables. RunAsyncRunner now fills the gaps when
parsing TASKSRV.
Fixes#119
