+
+
+
+**Scaler provides a simple, efficient and reliable way to perform distributed computing** using a centralized scheduler,
+with a stable and language agnostic protocol for client and worker communications.
+
+```python
+import math
+from scaler import Client
+
+with Client(address="tcp://127.0.0.1:2345") as client:
+ # Submits 100 tasks
+ futures = [
+ client.submit(math.sqrt, i)
+ for i in range(0, 100)
+ ]
+
+ # Collects the results and sums them
+ result = sum(future.result() for future in futures)
+
+ print(result) # 661.46
+```
+
+Scaler is a suitable Dask replacement, offering significantly better scheduling performance for jobs with a large number
+of lightweight tasks while improving on load balancing, messaging and deadlocks.
+
+## Features
+
+- Distributed computing on **multiple cores and multiple servers**
+- **Python** reference implementation, with **language agnostic messaging protocol** built on top of
+ [Cap'n Proto](https://capnproto.org/) and [ZeroMQ](https://zeromq.org)
+- **Graph** scheduling, which supports [Dask](https://www.dask.org)-like graph computing, optionally you
+ can use [GraphBLAS](https://graphblas.org) for very large graph tasks
+- **Automated load balancing**. automatically balances load from busy workers to idle workers and tries to keep workers
+ utilized as uniformly as possible
+- **Automated task recovery** from faulting workers who have died
+- Supports for **nested tasks**, tasks can themselves submit new tasks
+- `top`-like **monitoring tools**
+- GUI monitoring tool
+
+Scaler's scheduler can be run on PyPy, which can provide a performance boost
+
+## Installation
+
+```bash
+$ pip install scaler
+
+# or with graphblas and uvloop support
+$ pip install scaler[graphblas,uvloop]
+```
+
+## Quick Start
+
+Scaler operates around 3 components:
+
+- A **scheduler**, responsible for routing tasks to available computing resources
+- A set of **workers**, or cluster. Workers are independent computing units, each capable of executing a single task
+- **Clients** running inside applications, responsible for submitting tasks to the scheduler.
+
+### Start local scheduler and cluster at the same time in code
+
+A local scheduler and a local set of workers can be conveniently spawn using `SchedulerClusterCombo`:
+
+```python
+from scaler import SchedulerClusterCombo
+
+cluster = SchedulerClusterCombo(address="tcp://127.0.0.1:2345", n_workers=4)
+
+...
+
+cluster.shutdown()
+```
+
+This will start a scheduler with 4 task executing workers on port `2345`.
+
+### Setting up a computing cluster from the CLI
+
+The scheduler and workers can also be started from the command line with `scaler_scheduler` and `scaler_cluster`.
+
+First start the Scaler scheduler:
+
+```bash
+$ scaler_scheduler tcp://127.0.0.1:2345
+[INFO]2023-03-19 12:16:10-0400: logging to ('/dev/stdout',)
+[INFO]2023-03-19 12:16:10-0400: use event loop: 2
+[INFO]2023-03-19 12:16:10-0400: Scheduler: monitor address is ipc:///tmp/127.0.0.1_2345_monitor
+...
+```
+
+Then start a set of workers (a.k.a. a Scaler *cluster*) that connect to the previously started scheduler:
+
+```bash
+$ scaler_cluster -n 4 tcp://127.0.0.1:2345
+[INFO]2023-03-19 12:19:19-0400: logging to ('/dev/stdout',)
+[INFO]2023-03-19 12:19:19-0400: ClusterProcess: starting 4 workers, heartbeat_interval_seconds=2, object_retention_seconds=3600
+[INFO]2023-03-19 12:19:19-0400: Worker[0] started
+[INFO]2023-03-19 12:19:19-0400: Worker[1] started
+[INFO]2023-03-19 12:19:19-0400: Worker[2] started
+[INFO]2023-03-19 12:19:19-0400: Worker[3] started
+...
+```
+
+Multiple Scaler clusters can be connected to the same scheduler, providing distributed computation over multiple
+servers.
+
+`-h` lists the available options for the scheduler and the cluster executables:
+
+```bash
+$ scaler_scheduler -h
+$ scaler_cluster -h
+```
+
+### Submitting Python tasks using the Scaler client
+
+Knowing the scheduler address, you can connect and submit tasks from a client in your Python code:
+
+```python
+from scaler import Client
+
+
+def square(value: int):
+ return value * value
+
+
+with Client(address="tcp://127.0.0.1:2345") as client:
+ future = client.submit(square, 4)
+ print(future.result()) # 16
+```
+
+`Client.submit()` returns a standard Python future.
+
+## Graph computations
+
+Scaler also supports graph tasks, for example:
+
+```python
+from scaler import Client
+
+
+def inc(i):
+ return i + 1
+
+
+def add(a, b):
+ return a + b
+
+
+def minus(a, b):
+ return a - b
+
+
+graph = {
+ "a": 2,
+ "b": 2,
+ "c": (inc, "a"), # c = a + 1 = 2 + 1 = 3
+ "d": (add, "a", "b"), # d = a + b = 2 + 2 = 4
+ "e": (minus, "d", "c") # e = d - c = 4 - 3 = 1
+}
+
+with Client(address="tcp://127.0.0.1:2345") as client:
+ result = client.get(graph, keys=["e"])
+ print(result) # {"e": 1}
+```
+
+## Nested computations
+
+Scaler allows tasks to submit new tasks while being executed. Scaler also supports recursive task calls.
+
+```python
+from scaler import Client
+
+
+def fibonacci(clnt: Client, n: int):
+ if n == 0:
+ return 0
+ elif n == 1:
+ return 1
+ else:
+ a = clnt.submit(fibonacci, clnt, n - 1)
+ b = clnt.submit(fibonacci, clnt, n - 2)
+ return a.result() + b.result()
+
+
+with Client(address="tcp://127.0.0.1:2345") as client:
+ result = client.submit(fibonacci, client, 8).result()
+ print(result) # 21
+```
+
+## Performance
+
+### uvloop
+
+For better async performance, you can install uvloop (`pip install uvloop`) and supply `uvloop` for the CLI argument
+`--event-loop` or as a keyword argument for `event_loop` in Python code when initializing the scheduler.
+
+```bash
+scaler_scheduler --event-loop uvloop tcp://127.0.0.1:2345
+```
+
+```python
+from scaler import SchedulerClusterCombo
+
+scheduler = SchedulerClusterCombo(address="tcp://127.0.0.1:2345", event_loop="uvloop", n_workers=4)
+```
+
+## Monitoring
+
+### From the CLI
+
+Use `scaler_top` to connect to the scheduler's monitor address (printed by the scheduler on startup) to see
+diagnostics/metrics information about the scheduler and its workers.
+
+```bash
+$ scaler_top ipc:///tmp/127.0.0.1_2345_monitor
+```
+
+It will look similar to `top`, but provides information about the current Scaler setup:
+
+```bash
+scheduler | task_manager | scheduler_sent | scheduler_received
+ cpu 0.0% | unassigned 0 | ObjectResponse 24 | Heartbeat 183,109
+ rss 37.1 MiB | running 0 | TaskEcho 200,000 | ObjectRequest 24
+ | success 200,000 | Task 200,000 | Task 200,000
+ | failed 0 | TaskResult 200,000 | TaskResult 200,000
+ | canceled 0 | BalanceRequest 4 | BalanceResponse 4
+--------------------------------------------------------------------------------------------------
+Shortcuts: worker[n] cpu[c] rss[m] free[f] working[w] queued[q]
+
+Total 10 worker(s)
+ worker agt_cpu agt_rss [cpu] rss free sent queued | object_id_to_tasks
+W|Linux|15940|3c9409c0+ 0.0% 32.7m 0.0% 28.4m 1000 0 0 |
+W|Linux|15946|d6450641+ 0.0% 30.7m 0.0% 28.2m 1000 0 0 |
+W|Linux|15942|3ed56e89+ 0.0% 34.8m 0.0% 30.4m 1000 0 0 |
+W|Linux|15944|6e7d5b99+ 0.0% 30.8m 0.0% 28.2m 1000 0 0 |
+W|Linux|15945|33106447+ 0.0% 31.1m 0.0% 28.1m 1000 0 0 |
+W|Linux|15937|b031ce9a+ 0.0% 31.0m 0.0% 30.3m 1000 0 0 |
+W|Linux|15941|c4dcc2f3+ 0.0% 30.5m 0.0% 28.2m 1000 0 0 |
+W|Linux|15939|e1ab4340+ 0.0% 31.0m 0.0% 28.1m 1000 0 0 |
+W|Linux|15938|ed582770+ 0.0% 31.1m 0.0% 28.1m 1000 0 0 |
+W|Linux|15943|a7fe8b5e+ 0.0% 30.7m 0.0% 28.3m 1000 0 0 |
+```
+
+- scheduler section shows scheduler resource usage
+- task_manager section shows count for each task status
+- scheduler_sent section shows count for each type of messages scheduler sent
+- scheduler_received section shows count for each type of messages scheduler received
+- function_id_to_tasks section shows task count for each function used
+- worker section shows worker details, you can use shortcuts to sort by columns, the char * on column header show which
+ column is sorted right now
+ - agt_cpu/agt_rss means cpu/memory usage of worker agent
+ - cpu/rss means cpu/memory usage of worker
+ - free means number of free task slots for this worker
+ - sent means how many tasks scheduler sent to the worker
+ - queued means how many tasks worker received and queued
+
+### From the web UI
+
+`scaler_ui` provides a web monitoring interface for Scaler.
+
+```bash
+$ scaler_ui ipc:///tmp/127.0.0.1_2345_monitor --port 8081
+```
+
+This will open a web server on port `8081`.
+
+## Contributing
+
+Your contributions are at the core of making this a true open source project. Any contributions you make are **greatly
+appreciated**.
+
+We welcome you to:
+
+- Fix typos or touch up documentation
+- Share your opinions on [existing issues](https://github.com/citi/scaler/issues)
+- Help expand and improve our library by [opening a new issue](https://github.com/citi/scaler/issues/new)
+
+Please review our [community contribution guidelines](https://github.com/Citi/.github/blob/main/CONTRIBUTING.md) and
+[functional contribution guidelines](./CONTRIBUTING.md) to get started 👍.
+
+## Code of Conduct
+
+We are committed to making open source an enjoyable and respectful experience for our community. See
+[`CODE_OF_CONDUCT`](https://github.com/Citi/.github/blob/main/CODE_OF_CONDUCT.md) for more information.
+
+## License
+
+This project is distributed under the [Apache-2.0 License](https://www.apache.org/licenses/LICENSE-2.0). See
+[`LICENSE`](./LICENSE) for more information.
+
+## Contact
+
+If you have a query or require support with this project, [raise an issue](https://github.com/Citi/scaler/issues).
+Otherwise, reach out to [opensource@citi.com](mailto:opensource@citi.com).
diff --git a/__pycache__/check.cpython-311.pyc b/__pycache__/check.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..af4249eeebec0877e086ac9cda5d678a9a63a1ff
GIT binary patch
literal 2637
zcmb_d%WE4)7@yhaXm|DSn_7;vi5(+jTE|HX!3_mFp-sR^XkGFUZG`ZS9#oc(*R^SyrG
zeDm$-7r$QuWc=RzIlayS@F!jD457t5|BD6SHedh-8n_DqctC#J;{yQZE{%Y%$~VMj@9i5+O8PCaW3r`cu^xCnhqB
z3wR=(!=Gh@u33-7|EiBSL|41JuOL~xoNb=ca
zA%qU^D}P=YxA=aW@81Lv_S6qMlDsxr9bJvpV&&7+s>f4!nJ>$F!cRWGexsa+|5OyRt3ZxagFPvJUbp4&dsf_uqI5@E>X&H4HXGxI~RF=B;A
zHvy6DC*2?rGP^5h%RWa5m`vql89B1QqE}*NaYG5NovWTRwbhHYi<^Lhhdok0<4D2E
zQIlI8uZ@>aJEBzn*6He96)H?+#z9@?k(&Q08hV0;>YrIiwUOFH>ehzRvv$6EzCL0p
zQCo@9NUBHP5RPP5Q=X#m6BMq`SSV_vXcI+AxbC&t>TJE&QX;kzp^?0p(Kmz(noLc8
ziUyyc!TNCvMQjvlqR7?;;LChdh??gJA&&)A8>mg7I@^jNGu9M_$YX&+HaOG-hgyn3
zbD${k?{{=8;Ds%t<
literal 0
HcmV?d00001
diff --git a/benchmarks/dask_local_test.py b/benchmarks/dask_local_test.py
new file mode 100644
index 0000000..90c390b
--- /dev/null
+++ b/benchmarks/dask_local_test.py
@@ -0,0 +1,30 @@
+import random
+
+from dask.distributed import Client, LocalCluster
+
+from scaler.utility.logging.scoped_logger import ScopedLogger
+from scaler.utility.logging.utility import setup_logger
+
+
+def sleep_print(sec: int):
+ return sec * 1
+
+
+def main():
+ setup_logger()
+ tasks = [random.randint(0, 100) for _ in range(10000)]
+
+ cluster = LocalCluster(n_workers=10, threads_per_worker=2, memory_limit="100GB")
+ client = Client(address=cluster.scheduler_address)
+
+ with ScopedLogger(f"submit {len(tasks)} tasks"):
+ futures = [client.submit(sleep_print, i) for i in tasks]
+
+ with ScopedLogger(f"gather {len(futures)} results"):
+ results = [future.result() for future in futures]
+
+ assert results == tasks
+
+
+if __name__ == "__main__":
+ main()
diff --git a/benchmarks/dask_remote_test.py b/benchmarks/dask_remote_test.py
new file mode 100644
index 0000000..7ad8137
--- /dev/null
+++ b/benchmarks/dask_remote_test.py
@@ -0,0 +1,30 @@
+import random
+
+from dask.distributed import Client as DaskClient
+
+from scaler.utility.logging.scoped_logger import ScopedLogger
+from scaler.utility.logging.utility import setup_logger
+
+
+def sleep_print(sec: int):
+ return sec * 1
+
+
+def main():
+ setup_logger()
+
+ tasks = [random.randint(0, 100) for _ in range(10000)]
+
+ client = DaskClient("127.0.0.1:12345")
+
+ with ScopedLogger(f"submit {len(tasks)} tasks"):
+ futures = [client.submit(sleep_print, i) for i in tasks]
+
+ with ScopedLogger(f"gather {len(futures)} results"):
+ results = [future.result() for future in futures]
+
+ assert results == tasks
+
+
+if __name__ == "__main__":
+ main()
diff --git a/benchmarks/scaler_test.py b/benchmarks/scaler_test.py
new file mode 100644
index 0000000..d3a7d17
--- /dev/null
+++ b/benchmarks/scaler_test.py
@@ -0,0 +1,36 @@
+import random
+
+from scaler.client.client import Client
+from scaler.cluster.combo import SchedulerClusterCombo
+from scaler.utility.logging.scoped_logger import ScopedLogger
+from scaler.utility.logging.utility import setup_logger
+
+
+def sleep_print(sec: int):
+ return sec * 1
+
+
+def main():
+ setup_logger()
+
+ address = "tcp://127.0.0.1:2345"
+
+ cluster = SchedulerClusterCombo(address=address, n_workers=10, per_worker_queue_size=2, event_loop="builtin")
+ client = Client(address=address)
+
+ tasks = [random.randint(0, 101) for _ in range(10000)]
+
+ with ScopedLogger(f"Scaler submit {len(tasks)} tasks"):
+ futures = [client.submit(sleep_print, a) for a in tasks]
+
+ with ScopedLogger(f"Scaler gather {len(futures)} results"):
+ results = [future.result() for future in futures]
+
+ assert results == tasks
+
+ client.disconnect()
+ cluster.shutdown()
+
+
+if __name__ == "__main__":
+ main()
diff --git a/check.py b/check.py
new file mode 100644
index 0000000..31fb9a7
--- /dev/null
+++ b/check.py
@@ -0,0 +1,52 @@
+with open("scheduler") as f:
+ sch = f.read()
+
+with open("cluster") as f:
+ cluster = f.read()
+
+with open("shutff") as f:
+ shutff = f.read()
+
+# read_message(): read message ; HASH: -599833624
+# write_message(): writing message ; HASH: -1390068153
+def parse(s) -> tuple[str, str] | None:
+ if s.startswith("read_message"):
+ id = s.split("HASH: ")[1]
+ return (id, 0)
+ if s.startswith("write_message"):
+ id = s.split("HASH: ")[1]
+ return (id, 1)
+ return None
+
+scheduler = [[], []]
+cluster_ = [[], []]
+shutff_ = [[], []]
+
+for line in sch.split("\n"):
+ res = parse(line)
+ if res is not None:
+ id, op = res
+ scheduler[op].append(id)
+
+for line in cluster.split("\n"):
+ res = parse(line)
+ if res is not None:
+ id, op = res
+ cluster_[op].append(id)
+
+for line in shutff.split("\n"):
+ res = parse(line)
+ if res is not None:
+ id, op = res
+ shutff_[op].append(id)
+
+print(f"Scheduler: {scheduler}")
+print(f"Cluster: {cluster_}")
+print(f"Shutff: {shutff_}")
+
+all_read = scheduler[0] + cluster_[0] + shutff_[0]
+all_write = scheduler[1] + cluster_[1] + shutff_[1]
+
+# diff = all_read.difference(all_write)
+
+# print(f"Diff: {diff}")
diff --git a/docs/.gitignore b/docs/.gitignore
new file mode 100644
index 0000000..567609b
--- /dev/null
+++ b/docs/.gitignore
@@ -0,0 +1 @@
+build/
diff --git a/docs/Makefile b/docs/Makefile
new file mode 100644
index 0000000..d0c3cbf
--- /dev/null
+++ b/docs/Makefile
@@ -0,0 +1,20 @@
+# Minimal makefile for Sphinx documentation
+#
+
+# You can set these variables from the command line, and also
+# from the environment for the first two.
+SPHINXOPTS ?=
+SPHINXBUILD ?= sphinx-build
+SOURCEDIR = source
+BUILDDIR = build
+
+# Put it first so that "make" without argument is like "make help".
+help:
+ @$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
+
+.PHONY: help Makefile
+
+# Catch-all target: route all unknown targets to Sphinx using the new
+# "make mode" option. $(O) is meant as a shortcut for $(SPHINXOPTS).
+%: Makefile
+ @$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
diff --git a/docs/make.bat b/docs/make.bat
new file mode 100644
index 0000000..dc1312a
--- /dev/null
+++ b/docs/make.bat
@@ -0,0 +1,35 @@
+@ECHO OFF
+
+pushd %~dp0
+
+REM Command file for Sphinx documentation
+
+if "%SPHINXBUILD%" == "" (
+ set SPHINXBUILD=sphinx-build
+)
+set SOURCEDIR=source
+set BUILDDIR=build
+
+%SPHINXBUILD% >NUL 2>NUL
+if errorlevel 9009 (
+ echo.
+ echo.The 'sphinx-build' command was not found. Make sure you have Sphinx
+ echo.installed, then set the SPHINXBUILD environment variable to point
+ echo.to the full path of the 'sphinx-build' executable. Alternatively you
+ echo.may add the Sphinx directory to PATH.
+ echo.
+ echo.If you don't have Sphinx installed, grab it from
+ echo.https://www.sphinx-doc.org/
+ exit /b 1
+)
+
+if "%1" == "" goto help
+
+%SPHINXBUILD% -M %1 %SOURCEDIR% %BUILDDIR% %SPHINXOPTS% %O%
+goto end
+
+:help
+%SPHINXBUILD% -M help %SOURCEDIR% %BUILDDIR% %SPHINXOPTS% %O%
+
+:end
+popd
diff --git a/docs/requirements_docs.txt b/docs/requirements_docs.txt
new file mode 100644
index 0000000..85e44e0
--- /dev/null
+++ b/docs/requirements_docs.txt
@@ -0,0 +1,3 @@
+sphinx
+sphinx_rtd_theme
+sphinx-substitution-extensions
diff --git a/docs/source/_static/style.css b/docs/source/_static/style.css
new file mode 100644
index 0000000..e97cb2a
--- /dev/null
+++ b/docs/source/_static/style.css
@@ -0,0 +1,19 @@
+.wy-nav-content {
+ max-width: none;
+}
+/* Newlines (\a) and spaces (\20) before each parameter */
+/*.sig-param::before {*/
+/* content: "\a\20\20\20\20\20\20\20\20\20\20\20\20\20\20\20\20";*/
+/* white-space: pre;*/
+/*}*/
+
+/* Newline after the last parameter (so the closing bracket is on a new line) */
+/*dt em.sig-param:last-of-type::after {*/
+/* content: "\a";*/
+/* white-space: pre;*/
+/*}*/
+
+ /*To have blue background of width of the block (instead of width of content) */
+dl.class > dt:first-of-type {
+ display: block !important;
+}
diff --git a/docs/source/_templates/layout.html b/docs/source/_templates/layout.html
new file mode 100644
index 0000000..50f4d90
--- /dev/null
+++ b/docs/source/_templates/layout.html
@@ -0,0 +1,4 @@
+{% extends "!layout.html" %}
+{% block extrahead %}
+
+{% endblock %}
diff --git a/docs/source/conf.py b/docs/source/conf.py
new file mode 100644
index 0000000..f2fca81
--- /dev/null
+++ b/docs/source/conf.py
@@ -0,0 +1,74 @@
+# Configuration file for the Sphinx documentation builder.
+#
+# This file only contains a selection of the most common options. For a full
+# list see the documentation:
+# https://www.sphinx-doc.org/en/master/usage/configuration.html
+
+# -- Path setup --------------------------------------------------------------
+
+# If extensions (or modules to document with autodoc) are in another directory,
+# add these directories to sys.path here. If the directory is relative to the
+# documentation root, use os.path.abspath to make it absolute, like shown here.
+
+import os
+import sys
+
+sys.path.insert(0, os.path.abspath(os.path.join("..", "..")))
+
+
+# -- Project information -----------------------------------------------------
+
+project = "Scaler"
+author = "Citi"
+
+
+version = __import__("scaler").__version__
+release = f"{__import__('scaler').__version__}-py3-none-any"
+
+rst_prolog = f"""
+.. |version| replace:: {version}
+.. |release| replace:: {release}
+"""
+
+# -- General configuration ---------------------------------------------------
+
+# Add any Sphinx extension module names here, as strings. They can be
+# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
+# ones.
+extensions = [
+ "sphinx.ext.autodoc",
+ "sphinx.ext.doctest",
+ "sphinx_substitution_extensions",
+ "sphinx.ext.napoleon",
+ "sphinx.ext.autosectionlabel",
+]
+
+# Add any paths that contain templates here, relative to this directory.
+templates_path = ["_templates"]
+
+# List of patterns, relative to source directory, that match files and
+# directories to ignore when looking for source files.
+# This pattern also affects html_static_path and html_extra_path.
+exclude_patterns = []
+
+
+# -- Options for HTML output -------------------------------------------------
+
+# The theme to use for HTML and HTML Help pages. See the documentation for
+# a list of builtin themes.
+#
+# html_theme = "alabaster"
+
+html_theme = "sphinx_rtd_theme"
+
+# Add any paths that contain custom static files (such as style sheets) here,
+# relative to this directory. They are copied after the builtin static files,
+# so a file named "default.css" will overwrite the builtin "default.css".
+html_static_path = ["_static"]
+html_css_files = ["style.css"]
+
+# html_static_path = []
+# html_css_files = []
+
+
+# -- Extension configuration -------------------------------------------------
diff --git a/docs/source/index.rst b/docs/source/index.rst
new file mode 100644
index 0000000..30f03bf
--- /dev/null
+++ b/docs/source/index.rst
@@ -0,0 +1,31 @@
+.. scaler documentation master file, created by
+ sphinx-quickstart on Wed Feb 15 16:00:47 2023.
+ You can adapt this file completely to your liking, but it should at least
+ contain the root `toctree` directive.
+
+Welcome to scaler's documentation!
+================================================
+
+Scaler is a lightweight distributed computing Python framework. It allocates work over multiple workers (distributedly or locally through multiprocessing). The motivation behind Scaler was the difficulty of debugging Dask jobs that were hanging. It is meant to be a drop-in replacement for Dask Futures with improvements and additional features.
+
+Scaler's design focuses on simplicity, resulting in reduced overhead and debuggable errors. Based on our benchmarking results for ICAAP Wholesale Loss forecasting, it took Scaler Graph only a third of the time to finish compared to Dask.
+
+Lastly, Scaler's load-balancers work in a more predictable manner, increasing utilization of workers. The best part is that Scaler can be adopted in place of Dask with minimal code change!
+
+Key features include:
+
+ * Graph Task Submission - submit task with dependencies together
+ * Monitoring Dashboard - utilization of workers and task completion can be seen
+ * Task Profiling - seeing execution time of functions
+ * Logging of Distributed Work - send logs back to the Client from workers (under development)
+
+
+Content
+=======
+
+.. toctree::
+ :maxdepth: 2
+
+ tutorials/quickstart
+ tutorials/features
+ tutorials/configuration
diff --git a/docs/source/tutorials/configuration.rst b/docs/source/tutorials/configuration.rst
new file mode 100644
index 0000000..6110f54
--- /dev/null
+++ b/docs/source/tutorials/configuration.rst
@@ -0,0 +1,91 @@
+Configuration
+=============
+
+Scaler has some settings that can be tuned for performance, but ideally users should not have to concern themselves with these because the defaults will give decent performance. A full list of the settings can be found using `-h` flag in the CLI (as detailed below). List in this page are the expected settings that users may want to configure.
+
+Scheduler Settings
+------------------
+
+For the full list, use the CLI command:
+
+.. code:: bash
+
+ scaler_scheduler -h
+
+**Protected Mode**
+
+.. _protected:
+
+The Scheduler is started in protected mode by default, which means that it can not be shut down by the Client. This is because multiple Clients can connect to a long-running Scheduler, and calling `Client.shutdown()` may inadverdently kill work that another user is executing. This is also because Scaler encourages strong decoupling between Client, Scheduler and Workers. To turn off protected mode, start the scheduler with:
+
+.. code:: bash
+
+ scaler_scheduler tcp://127.0.0.1:8516 -p False
+
+Or if using the programmatic API, pass ``protected=True```:
+
+.. code:: python
+
+ from scaler import SchedulerClusterCombo
+
+ cluster = SchedulerClusterCombo(
+ address=f"tcp://127.0.0.1:{port}",
+ n_workers=2,
+ protected=protected,
+ )
+
+**Event Loop**
+
+Scaler supports ``uvloop`` as the event loop for the backend. This will provide speedups when running Scaler. ``Uvloop`` needs to be installed separately:
+
+.. code:: bash
+
+ pip install uvloop
+
+``uvloop`` can be set when starting the scheduler through the CLI:
+
+.. code:: bash
+
+ scaler_scheduler tcp://127.0.0.1:8516 -e uvloop
+
+.. code:: python
+
+ from scaler import SchedulerClusterCombo
+
+ cluster = SchedulerClusterCombo(
+ address=f"tcp://127.0.0.1:{port}",
+ n_workers=2,
+ event_loop="uvloop"
+ )
+
+Worker Settings
+---------------
+
+For the full list, use the CLI command:
+
+.. code:: bash
+
+ scaler_cluster -h
+
+**Death Timeout**
+
+Workers are spun up with a ``death_timeout_seconds``, which indicates how long the worker will stay alive without being connected to a Scheduler. The default setting is 300 seconds. This is intended for the workers to clean up if the Scheduler crashes.
+
+This can be set using the CLI:
+
+.. code:: bash
+
+ scaler_cluster -n 10 tcp://127.0.0.1:8516 -ds 300
+
+Through the programmatic API:
+
+.. code:: python
+
+ from scaler import SchedulerClusterCombo
+
+ cluster = SchedulerClusterCombo(
+ address=f"tcp://127.0.0.1:8516",
+ n_workers=2,
+ death_timeout_seconds=300
+ )
+
diff --git a/docs/source/tutorials/features.rst b/docs/source/tutorials/features.rst
new file mode 100644
index 0000000..76ba296
--- /dev/null
+++ b/docs/source/tutorials/features.rst
@@ -0,0 +1,257 @@
+Additional Features
+===================
+
+These features are not needed for standard usage, but are helpful for monitoring and other use cases.
+
+Scaler Top (Monitoring)
+-----------------------
+
+The scheduler has an address that can be monitored with an ipc connection. The exact address will be in the logs
+when the scheduler is spun up. Connect to it with the `scaler_top` CLI command.
+
+.. code:: bash
+
+ scaler_top ipc:///tmp/0.0.0.0_8516_monitor
+
+Which will show something similar to top command, but it's for getting status of the scaled system:
+
+.. code:: console
+
+ scheduler | task_manager | scheduler_sent | scheduler_received
+ cpu 0.0% | unassigned 0 | HeartbeatEcho 283,701 | Heartbeat 283,701
+ rss 130.1m | running 0 | ObjectResponse 233 | ObjectRequest 215
+ | success 53,704 | TaskEcho 53,780 | Task 53,764
+ | failed 14 | Task 54,660 | TaskResult 53,794
+ | canceled 48 | TaskResult 53,766 | DisconnectRequest 21
+ | not_found 14 | ObjectRequest 366 | TaskCancel 60
+ | DisconnectResponse 21 | BalanceResponse 15
+ | TaskCancel 62 | GraphTask 6
+ | BalanceRequest 15 |
+ | GraphTaskResult 6 |
+ -------------------------------------------------------------------------------------------------
+ Shortcuts: worker[n] agt_cpu[C] agt_rss[M] cpu[c] rss[m] free[f] sent[w] queued[d] lag[l]
+
+ Total 7 worker(s)
+ worker agt_cpu agt_rss [cpu] rss free sent queued lag ITL | client_manager
+ 2732890|sd-1e7d-dfba|d26+ 0.5% 111.8m 0.5% 113.3m 1000 0 0 0.7ms 100 |
+ 2732885|sd-1e7d-dfba|56b+ 0.0% 111.0m 0.5% 111.2m 1000 0 0 0.7ms 100 | func_to_num_tasks
+ 2732888|sd-1e7d-dfba|108+ 0.0% 111.7m 0.5% 111.0m 1000 0 0 0.6ms 100 |
+ 2732891|sd-1e7d-dfba|149+ 0.0% 113.0m 0.0% 112.2m 1000 0 0 0.9ms 100 |
+ 2732889|sd-1e7d-dfba|211+ 0.5% 111.7m 0.0% 111.2m 1000 0 0 1ms 100 |
+ 2732887|sd-1e7d-dfba|e48+ 0.5% 112.6m 0.0% 111.0m 1000 0 0 0.9ms 100 |
+ 2732886|sd-1e7d-dfba|345+ 0.0% 111.5m 0.0% 112.8m 1000 0 0 0.8ms 100 |
+
+
+* scheduler section is showing how much resources the scheduler used
+* task_manager section shows count for each task status
+* scheduler_sent section shows count for each type of messages scheduler sent
+* scheduler_received section shows count for each type of messages scheduler received
+* object_id_to_tasks section shows task count for each object used
+* worker section shows worker details, you can use shortcuts to sort by columns, the char * on column header show which
+ column is sorted right now
+ * agt_cpu/agt_rss means cpu/memory usage of worker agent
+ * cpu/rss means cpu/memory usage of worker
+ * free means number of free task slots for this worker
+ * sent means how many tasks scheduler sent to the worker
+ * queued means how many tasks worker received and queued
+ * lag means the latency between scheduler and worker
+ * ITL means debug bits information, I means processor initialized, T means have a task or not, L means task lock
+
+
+Task Profiling
+--------------
+
+To get the execution time of a task, submit it with profiling turned on. ``.result()`` needs to be called on the Future first so that execution is complete.
+
+.. code:: python
+
+ from scaler import Client
+
+ def calculate(sec: int):
+ return sec * 1
+
+ client = Client(address="tcp://127.0.0.1:2345")
+ fut = client.submit(calculate, 1, profiling=True)
+ fut.result()
+
+ # contains task run duration time in microseconds
+ fut.profiling_info().duration_us
+
+ # contains the peak memory usage in bytes for that function, this memory peak is sampled every second
+ fut.profiling_info().peak_memory
+
+
+Send Object
+-----------
+
+Scaler can send objects to the workers. This is useful for sending large objects that are needed for the tasks, and
+reuse it over and over again
+
+- The object is sent to the workers only once
+- the scaler API will returns an reference that link to the object
+- on the workers side, workers can use this reference to access the object, but this reference must be in the
+ positional argument level, **it cannot be nested to the other reference or inside of list etc.**
+- This ``client.send_object``, objects are still get serialized and deserialized by Serializer, if you have special
+ needs to customize your serializer, please refer below :ref:`Custom Serializer` section
+
+.. code:: python
+
+ from scaler import Client
+
+ def add(a, b):
+ return a + b
+
+ client = Client(address="tcp://127.0.0.1:2345")
+ ref = client.send_object("large_object", [1, 2, 3, 4, 5])
+
+ fut = client.submit(add, ref, [6])
+ assert fut.result() == [1, 2, 3, 4, 5, 6]
+
+ # this will not work, scaler doesn't do deep resolving
+ # fut = client.submit(add, [ref], [6])
+
+
+Graph Submission
+----------------
+
+For tasks that are dependent on the output of other tasks, they can all be submitted together as a graph. Scaler will handle executing the dependencies in the right order.
+
+.. code:: python
+
+ from scaler import Client
+
+ def inc(i):
+ return i + 1
+
+ def add(a, b):
+ return a + b
+
+ def minus(a, b):
+ return a - b
+
+ graph = {
+ "a": 2,
+ "b": 2,
+ "c": (inc, "a"), # c = a + 1 = 2 + 1 = 3
+ "d": (add, "a", "b"), # d = a + b = 2 + 2 = 4
+ "e": (minus, "d", "c") # e = d - c = 4 - 3 = 1
+ }
+
+ client = Client(address="tcp://127.0.0.1:2345")
+ futures = client.submit_graph(graph, keys=["e"])
+
+ print(futures[0].result())
+
+
+Client Shutdown
+---------------
+
+By default, the Scheduler is running in protected mode. For more information, check the :ref:`protected ` section. If the Scheduler is not in protected mode, the Client can shutdown the Cluster by calling ``client.shutdown()``. This needs to be specifically enabled when spinning up the Scheduler.
+
+.. code:: python
+
+ from scaler import Client
+
+ client = Client(address="tcp://127.0.0.1:2345")
+ client.shutdown()
+
+Custom Serializer
+-----------------
+Scaler uses cloudpickle by default for serialization. You can use a custom serializer by passing it to the Client.
+
+The serializer API has only two methods: ``serialize`` and ``deserialize``, and these are responsible for
+
+- function
+- function arguments
+- function results
+
+
+**All libraries used for serialization must be installed on workers.**
+
+
+.. py:function:: serialize(obj: Any) -> bytes
+
+ :param obj: the object to be serialized, can be function object, argument object, or function result object
+ :return: serialized bytes of the object
+
+Serialize the object to bytes, this serialization method is called for function object and EACH argument
+object and function result object, for example:
+
+
+.. code:: python
+
+ def add(a, b):
+ return a + b
+
+ client.submit(add, 1, 2)
+
+
+``serialize`` will initially be called three times: once for ``add``, once for ``1``, and once for ``2``.
+The result of the ``a+b`` calculation will then be serialized and sent back to the client.
+The client will ``deserialize`` the result.
+
+
+.. py:function:: deserialize(payload: bytes) -> Any
+
+ :param payload: the serialized bytes of the object, can be function object, argument object, or function result object
+ :return: any deserialized object
+
+Deserialize the bytes to the original object, this de-serialize method is used to deserialize the function
+object bytes and EACH serialized argument and serialized function result.
+
+
+Below is an example implementation of customized serializer that deal with different types, but as you introduce
+different types, you may need an enum in front of serialized bytes to indicate how to deserialize on other ends, for
+example
+
+- following example will serialize and deserialize the pd.DataFrame and Integer specially (not use cloudpickle)
+- all other objects will be serialized by cloudpickle still
+
+
+.. code:: python
+
+ import enum
+ import pickle
+ import struct
+ from io import BytesIO
+ from typing import Any
+
+ import pandas as pd
+ from cloudpickle import cloudpickle
+
+ from scaler import Serializer
+
+
+ class ObjType(enum.Enum):
+ General = b"G"
+ Integer = b"I"
+ DataFrame = b"D"
+
+
+ class CustomSerializer(Serializer):
+ @staticmethod
+ def serialize(obj: Any) -> bytes:
+ if isinstance(obj, pd.DataFrame):
+ buf = BytesIO()
+ obj.to_parquet(buf)
+ return ObjType.DataFrame.value + buf.getvalue()
+
+ if isinstance(obj, int):
+ return ObjType.Integer.value + struct.pack("I", obj)
+
+ return ObjType.General.value + cloudpickle.dumps(obj, protocol=pickle.HIGHEST_PROTOCOL)
+
+ @staticmethod
+ def deserialize(payload: bytes) -> Any:
+ obj_type = ObjType(payload[0])
+ payload = payload[1:]
+
+ if obj_type == ObjType.DataFrame:
+ buf = BytesIO(payload)
+ return pd.read_parquet(buf)
+
+ if obj_type == ObjType.Integer:
+ return struct.unpack("I", payload)[0]
+
+ return cloudpickle.loads(payload)
+
diff --git a/docs/source/tutorials/images/architecture.png b/docs/source/tutorials/images/architecture.png
new file mode 100644
index 0000000000000000000000000000000000000000..bf7c8f160843823b1c7011741a3a7099960c002a
GIT binary patch
literal 36570
zcmeGFcUY6x8$XVNG+IHSRunC()`0^->i}eFQE`Hbh|FkFP(~6#2oNBORcevpA}ULz
zh>8qZ34|n!3TOmGgh)0cVJ4B0gb;p68r%pL^f0`#g^~
zk2yIm*8E9RO-*g_!2^3wsHthQ)zl!>-)n$RHZDoq3%;p`o^bq4t)SUp0Q~Zu|DL0J
z)YM)lYfYVnfNRKOM^5eopYPtS`1uTLlj|V(y3U;ozOM1?U4NRj+N)=sJ8QLf_gatM
zU%b1)hrhUYtpOi;bm=VL4zB*<*|pZQYqb}+fxQO&;L*Lt3;fWtR{2BETF-89&&=oG
zPLFQw<-e}+>RRmyrs`Si1-^E#@#t3m>M{Lcm+HeF@FBPXTm_eOz!1;wuQ!9)l%r;^
zs(x0@qx^8@@5~VIZWEs#lQTUVeS1yMaW?yL%rA1c1or(J#N7tx?g;6#xZH1r9I%cW
zv_%c>zB05YZrI`4@ctV-$3OW8{^B1@5FAPr98D28-4Qt7{c`O7m*bd`laEH-GR8cf
zj(g__eVz%S&qZfmh|U#EoPRyxUo5^*IuU@C1Xf6bD<=^IDS{*osh$e0or-%ZD5T4(
znna(5<*YvGkVpakEZr(t30S}X@(CvwweP-P^n>mSgP+%|H{G(`(q{MGeFu&lIdxE=NS8uUv~yNWP0n%gWBpD=fj{38cCPa!VV9#_H)C5{!wZ`<=>Pff*Mf
z4!DP^sr_25{HMPDoCi%!?bf4%d-t4-@EMj#uF-E$4gGxbKOG8Qy?BT2!JGFE9C~_y
zCnZLg-d*WzfFd60?hUo^HCc(O_e{^rBTE~cmmyx_HLnMRTl|z5+uAZ#NAP|QN%^!O
z!a4ip7ZWv0C|{#&4IXHb<9@%#f1T4)Q|oVRTy*mG@H{m&!<7%OZoK2IruKOC>TSpN
z6>F)fT|Rei{>ua5>S|Y#ZoS{V?5&xa+K1b>S3dux6R6~_Uw_qb{WjS*o_$k|Z|3yP
zYy5xSIX?VJK84*6yJWX^CIB?Yk)}9>XxY$Hl3ZwxmV_=3TaR0hQ&THCyx(sfHmH0k
z${Gf(@wU6=?OnI;G>!*^~!eEUjBBfat%%u?K0
ztEIo#t!7^hMD2?AufJ})Pl%OLn*OR}p(y(N@S3JOBrZ3Cx|Cc|b8mLWrJVw-dl6ia
z|E_{i>WvUsn4qx4#KGPTIRE{##Qd@#X5r||se&+8W*kb2Gjy=$&Z(BVR)V3wI
z`v*&%t6?mf8^ZZ!9LjGt$USSlSI@|L+$3Sh@XNe(z0}k`uU)bw1o!*aUoaqrEtv6ndo_Cc~wJf`Oz@!&A?j=kgZ!d{o--Awspw$IQBI!q5XgH9VUlG+vipGApmp$^DyfO>sO5&ko^=z#4rvi^OOsOFP9j=f&AloTwdMFpsFc-MESAE8Jpd{-YlBX#
z(-3S0LtM;Tt>cNX_C8f(Ydx#Xxp4m)o9b1XeEvG>atgtbdo`o*fk%JLb%r_l;^7%V
zq!UlpXNrFmbhvE8L$Kn-^@}0Mbzf&Nzdv#yW8?z_i6so;ONuw*{Jp+vLhyN;fjAnI
zd9-G;`M;8H<%z!+JYzp)&r_>5Tlw(eZsw6H)=^`pf6egf<@!sSd?>OL^>>`7>+E!z
z!TvV5?J(k1$qx{u7`q<19_Ko{$!8zJ>ox5EkYY%(^D@4J9^GO4aR~P1EQ@qc{G=0*
zNXAn(@>glXj7Gl{7Es;YQG(Bs3r^m%^Zyup(dO|j?&V>ohVH+ncL2@s)(=%2*Mw!3
zl%H8JXVUG|4&2zLC)Cl(aPe79ezQ)OJaX(+!5oTz%5%ZV;Om`8-gxmAh(&Uz-E}F2
zBnykN)r7guD$&g#T=n~9;-rjZ?x(L@C?3pUY%4%_4q@DDt;iN(Wx;
zpkF7M^)}S+54Am4!$92~t%WK^F=gdd%)`m25n`Whg_t+##96BU*s0Hd)_n6LFga6t
zxje+t6vxP?hh1E~A2VZQ4?j^?QWooDgzme7Y8WrG6PqyBzt=FR2NokY&aQfq7qdT{
z0mbJFzlUH2%d8W}pF^-Hj|kL|rQz&7FJzHJ-(7@WwTZ4_M7_yhTHt7g%d`H;(*$F+
zWA>hH>=vbqgSdcj1a;~i?;nKd6(WBNE0sm{y&2-{boo|=d@HuT6{z3p@Nd|~Y;gYu
z(fofEiK>oFS@*Y=x|-TC&%yjSfTQgWet}iiH)>@$%#EPHQu~w|l?O7cZ#R77!#92S
zzt%2Vv$LX&USP$7K2cwm{lf;Zk?Kw{EPz*@Jo-KOWj*^Xw9OxDgK-4E?1tc-x^`Vv
zdV>vLf7&d|MPyV1F#YiZxeLxofr3E9hoz?byKG=n)KGXv=Aq{_2E^*=6oHyiSe-(dquajmSUum2}#2*C+&IO=Q&N
z<$UBzy0Wo8(Kg6=_wG%{w<~8e59SBUM!VW`Ll*4zo=JjNaD^vn*I9N^O_&4fB`LGF
z!v1D36$7H{rt_Yg&2(|+Knr(5IrD=Z+#Bcq#Mby5r>zj7zsonTyC(}*i
zZ*JwAgM?f;`+sRXNmtjAmybKb5(7TGjD=@TVd>pUOn27D&!3Pb+drzEVAou;zwyZh
zf^h&nlG&(gKoOuz!{9B~d@WA?=ZbTsPs7%P>^Slh&|aFIEU8CaZX%knO}6~;Kesz5
z2524EY`E)ny|mYMe&uH9UaGf#7hpTB)QEyWVjgK}HH)hqjBCPHKaCpXRy}w(i-Mb+Y&%8dxwq*Qwt?`sNl<9vq{j_H}orPBP)I9aJ
zi}(Tk!N+kZ9MAoNLgF;8^Ie~^
zNmw_X9dVzh?trbrYHw)YQn?mE3X?PCEAZAZ;=78n5*Ja`XZg!h<(R!C(LGDw;I;U5
z<+{zhDq0U`W`u+hgkAl#9andh5
z_hobfdN)LDIhT)RgBRipFQD>ajL9}P!;-$MF@Xap0`CS*@q3;tkLL4Z(`cP8-6647VBi5khYrh>6;ja6r9CwPu8|);itb!UHGaxA
z@np?0Nv{{=E`xavq0yx*iLTVJ&^r}nWOLdkh{t<_Pa7J&D#cFz
z#5pk*m6s;zTBF6MJ_?6B<0BMK5b^1`dnlLRY}=Ph{^%`wImpxy2QYmg!<%AJP>c2#
zad~BenD=3YF5GZmRQ`}>%pHgrK`RkCPkcc84&R%znaZ;A^2=x5GKSEgr8`)FB8#9!
zxlXW5P4CV0g6H~q7cfDoJ!K`eCT<9w3R{%o4}H&nmdvrwq_d=yg5j#Ep+PzQmeUJ5
zijZiA5BOWZFg+e-hX_2o{BlR-#%vRJgbvm=Z|x*+%Nz^oIc>eNvaqn{*^l8UM#u;?
z=W=*E$+S5?$FPt>td@V|a>LW;fjE;>%BAzrwfUR756odeL$P>a=TYkdMdhXDl94k#
zxAXMKO|}EAfjUih=+w50nY5OK_P?B7v5jCT(J6Y#@}!^Vh~Wr+Ba9Er>9>7F;k`sh
zPI@#1=>?IPeusnM!gRAHEm&1G*M+j3Y%Zf=YID4EkXEcwbWD@K8Hs
z)#`}$?Ilk&0>b#aUZ@P_7{Yn)zqaCsV_C1YNYJCwY&nsE(YW6@7bzm_=lMvbK+2qJa@|rcP&fj_`ZVyC^{2=Z5iF>OZCz24|5m3*gmln
z&h3Ex$L$;szO9dE+kMW7`Mf7M`Ce5?rHSMex%(-v4
z#QS$~aACmi8rxox>3W4RAaw_J6k7^inL`N!P6|JMmzxXEM{PvP2%txZN%$GCh1@Mx_jE
zHyXC$W%zONT3U|ICR{w-2qHc=Hzcp`GsRjnV$oiLOOX+Mt?rcpSFDFvBEzP*{06R2
zlqwQ86Jqg_@#gx{;xA{F9B3QUsogqviVKd1z38uZRnRC4E%1JmEM{?r5H3w(ZflGr
z$he*O4gzoJRX|yEOo?0J#0TXzI%Up{$LCHFK>>@B39xB#8OK
zPPP|OSy2p{yJC`mVfl_-g%s3}VP{7k6OKlIa6^cpzLL0$wFwJ$D5OOy@+AKuub>TK
z5sbl=6;#0T6}(sI804@AD)+RwK)B~!kiRP-xyfoEemUJCxTP8WThd%TUqlk`tgVkS
z_8N#8kLpwsAc*K6+7V?`)}KxQJ>}P=z9%}Haq-klIr@jB^>Zx56eZ|klxEimb-NRP
zsVb54aK5~l>P@z9qK1x>Ltn}uNPM=LkKhAD`~&U(;DL+t8{5aEFr%2yli0DOYf$qJ
z;zYm`f+;U4jyAw->%HG+(Vo4UK
z@8KN8^47)9!H#HAC(XPsbFn!|79P861H3T*K#k$Ue|rIEj*w4LMmobW-a9`m<;9wt
zyOmN9%jmZ^{M%rtFAWN#?OG=05wBm{ZjPf|{68<-?o9A;+3j3BwZjAt;`HXKc{M
zzSmT6+Iaqd_+rZl>C{W6SYcYy9=zJApQR=9q=(?2cQ
zJyyTH`O(B^@lG*;>r)d9Et5XgK~@K(-RDQVZMWZxf{yu)wGGc{@+f5x1HtBBuXr8y
zSJ~!!Bkn|Kh3yc&z?*3!Eom=<(|+xwe;!JYlnC7z7apZ0!?teaO3&Ji&sF9+3`r)P{&}Z0-SR=
zIrGfW3Ktjt=-3TzlIvT#DJ~-b_u*d4H7)+jK)xx?v#^qdNOnhHeJ&A*Fvbh^ynA=1
zn_4SeXuPQHIiao_EwuQf^{g1tXt7Uu79u#0dliBm>7>C)jE;`XI9kCM9tIkVmz^8-
z=(O>xX=-|Nm|IuQz<7NomB_N_w%h(!E{r8wu(rkGGg+qKinm0hf$jR+QGJ}
z4f!Z%%OZAps)co4F!bEmTCvaH6=jNfD9;qywmCgAu%RKap=Uo7OfvV^@WRP&G8R}g
zmD|E}KQZI*qEgX!fiQzYcDea$MM@%OnAusfFch4uoo%tr3}--)J@vI@{r1pEe|Sw(
zVu#=PVB?~ZSAMO7>J1xl498Nf(T!wd3;rN(d{LOce
z|6?p*TR>!+0zw0i&u8(|HG;Q`d;2WCYMNR$4JBJT)z~6(Vm1xaMl0rS2^f3%OXi6-
zb+??5k{@>NdoAqr<9S_3<@Q+?1)ABuf&&`O^u{mx_(|$cVZ34241&z_Ff+b=U6a52f0Q842U-(s9Rv$<`}7Q25VVZ!
zbo3-5_t_i~wG|Hvm(j~>xci~|Vx%R0XMG9F!{O#f+z|G(4jccW*29`+yVp)9sKDA)
z6Q;tqf5>zs5C(C9oY2S;m!-*2#Mqxbx^#`9!RBTB=>5U}Lj>>;VZj?Yu~&y*Jinck
z=BY*Y%y{&9Y?v2Ei;)N&4X|txeDpU#2KhW#YT4JNz%ZSlabm%{l;Te7Y)KDh>>~dVZ{rn|Tj2YZ8j~$*_dg-rl
zsho2sUXSi{k=cgyFF8^JdH&0EKX=Qs^LezF1;HfmO;^&E33-Vomz@fy2O7P8eL*r-
zpRlvVzKAU{jf_3@{phs^$97H@I)#)UJ29o=Jo9;)imE}$;Zp#D;($wxfZ_TP7Jn|5WGRLc~myKfG
zkJI9Z84OV){qzMsj2{yzd?569gx4?%F}wtR*4bb*F%#*`POFJW#;yM8?tsZZS6ye^ZH+&
zB39x2+m~BpGN@*_tR2adK_mYVLK`cPFJjtXl@MMRmEYzSC${s0`>BE9)**Nzx7iOI
zc2`CqmyaYLgS$zeFDb|hkz%TFdCR0l#X;3xQrlrPkF&wsg&LZ2PsCaCDhEt-b00+$ejfDX}PYI(L_a;`;(+lIaVPj!Lml-C4a=-
z@xHRPj01DQGQ&3whFiQhOys_96)7LhXePrl9ROrQ_dz#`0`#0cptz#rQ0yrR?2YMk
zWPA3v;z7^Hu3dbROz$FAOR<%_9#rhWgY!kEtd=xXN7$yp{g;Mq8MFi5!*-}Ag=|>W
z@Ts>x=6P8JIFRS)8R0{gGh_uWKGqPT0Yt-M9?YL
zC@u$`o^ujl#p(Ot*sO;#Jo7htteY#`HTF-@!z`h_u-aNL$}Md%58YAcb7{s0y0qCh
zUELrbAPoy}3cj)YpwSY=QU2qn{~}gLj-sP^b73;iDWf|Hh2Uq
zJ<~BOhL;Mj?;bJ)zyxH6EfAu$mGsgrQ9AH
zFu=*5DSnxoxFxH=+rD3_R~#hxj!#`O^uoOOTvwl|=k_jtpO~@l-6*`VHVhK(BdR~I
zw%)M@!p5ayN(@h~Ma#I^G&kr);+_D8JS~~UvVK^JoQLG;^0SO00}EZz_xDN7sVb+u|KHBy(O6{a)Bt7ae%%-Eh2fuc(%~
zkS=t=p(oo`3rbB=;2e=zf|Hz!%32iFGnx#g6D~=*sr%K{{FUrjwH?ad-fMGV>~)7j
zd7T@^c|szj2fWIv<{6Y^N~0?J55a8TyEu>*4#``ZI{R`*qtvQbg8T@PVq!*&?ct}8
zJLMB^7FhV8V|nRAb+-5js)@b;%Cx+SkJLn7>f^EaEc6Bo?SY#K7&zTe2|&H$(%n)HC1Du1Y`dh?RjwxjAV=+#Y-_IwpNU#IA6oERR8=11Y@aO
zWfGPOZM7@$E(rCA$@_s|8T^GX%Kj>!;f{rLo(s<8rL3S|Q1a4O($!{7F*oGhZBVN|
zqiI$x5jJisBwF={l#^n|T6qjG`&OC>!kJS9E+ll@_udo9r1mjk0XH;7BW)TgzvpjBj%lb;_v=7B!fYhqarEYShP|<&$-uZqTTWg$t?|zC)V8
zqw`Ouaykw!r|*lD2lb2*(&t0;9#=*4T1Q%pb^@V~8P9L3XOgmL{+&zp7R2UKMhh4F
z57_LgHYkN38cQ5J?eWJc*w!SdZQu!UgZ^gYZ4nW
zfxmx)e!L23a#w;iusQjn%)I%}{X4x;8~xi5ZVA(@j59*Fm2OLtMKTd=cFRdncQPLMq`
z{K-nM^E(taLM@ohuj8JDn7$qXBRc!Db(8wDk5vwmC+)(1h?pY^smzKgm%PGQ*irEt
zT0WCJT5n)5o>9LVN^L--Yksa}?p3j4IHWv_sOSjXHu)2C%D3qmyTC?Lt8XD#Y-PoY
z))JMuWdd)0<(jGW`4$s|^-fKgGc7n^l$uv+yYrOZ&jCiIPrR9@(e?Gq5>m>X+%_6)r6BT+RtL?o#b668*Bd6yKT_ak)lHIL0b?VT#^>}lVi8sQ<
zQCzBH_naNh_oO`rB={Vwu{DF}vER=+ld1Vnx1p_QzGQoj7ufaa`@95J+#Vp`5i(xj
zMeGN@UZ2W+ppQ4qXNOCJjbe7!c-wc;KErWGYrGkSY$HsZ)|QHWxhsLw;CP-S1HwL!
z>R6A(;ftf=Cm*-?uoRpj4*X4!f~k(!dvqXr0%-@PGJ{Wna?l)Zk(KU!T3&@`b2D3*
zu@a8ZWV@(R;14GedU`5s(!tyrGf9Tb&br%of-|H8cLf4|&B>Ca>(V1Yh5!O>G%yT|
z;52%u%%5;+rKRA}F3Fd%r{KgxlT4aZaq+oShgDTl=2ec}l2zZ7?^+WRv9RxF;4i*2
zLcxHosh4s33i+-C<+R4K=w!Q;Q)O=1x
zMmaU;A8K)8VgTs8THMwE!7z7Vqz_8EBu+3c8M3{RMXW;g2>V*jAAL)@8tRU41IYuY
zBVOq}$3Vb+q@Z=GmnR-aq*30t95#pBgh%>=OnkG4*=fRWo`}$EJwT6~-Yf;Ff_BC7
z_sLLGbAM@s58vh;>CaGaM9TLd0tTS=`_)vtJ-zw&Wg1SZo#m+=Q`>
zn08pZ;<^E-&Q?(FI*?5Jm5$9@y03;I>MtoINSG!{V(R*k1tj@Yq#VBx0lNk0(a*SB
zDlHdy!5mtw7_Xjm)+SDpw}MEu+8y|Ux@#V{
zeZ8tnD<71!g2!lY;A2=J2mP0J7<~#lD_z~7w6anV1H+}+>UzJ*RPHS+a*erTFElw0
zPhP5L&caDKDHdjm$wa8_nR0DAT4mn>2$uJE0wxCU;;G+cE2(H53mM7>Pd+aJayzIb
zhOvrunFiEtnVuPAl_FJ_9v?AWL19Lr8_(ZQ!@kKH(}~vDw*3%5A3jV5<@M}5JQkX5
z+eOPJcur4phUd6y!S=rZS7J(db-mEkIMe~2^}-ufmn6C7!0dafy!=)^Ae4J^>tT+Y
za6^lct$r5vyJ@EHS{6gZ+W}7+#GgYk%qT^h(oH?xW8fsu*no2!>(#=*nTDe>2FSTU
zdPebyr0Qd?c6Yq&RZlvNu+9(5Zh>tE?fGkQT8Kq+{G6n_qDrZP&5NtG;1vtqN*
z*BeGN1Y`3L1y1&KNdLqqc_H)?>;io%U;A9ZV`+V{9bb2J+|+#cuKvrPlcCXwy7TTy
zYZddC(N`Hoa_
z^npn$Hfi#2f-tVlc?)tTr9TB)0u7lAyfT$=65(8XU9Z42HfsS$5kr4%
z@=^Cc*=CM)?%Me#q}>#pq_KGSrT8N^et+K|g3%5xHHyiq3ct8&dY7A$ZN>(RKu$W@
zJP0{7D)6eY4R$767Iwr#bx=K{gfuswFU_daVV;Mx#8{=tWq5(qEVobt?%I=80oK`6yy7Y^-
zqTxXJ;>qWeGARo^DHk4YS-51&O~Rc@7>O0G1w$o;HPzEnuzbe)=R;!t4mwXL$On7?
zE(YNQg7tk*@r`KvVIAOw?Zl^jlG-qeA2j)$U&c47l9+j|V)W!Fvx#h1shwWlo_0r^
zAG0nTfMnLgQu=bN&$}lI1k5#(1Ghc=q(RvWEY1DAVVK_lC@Yuo(
zExsu}pq9(wVC@4vximPugBBaSS{~Hks}r3sdXa$|9iX*Pbp#=CX7taBg6{!9mjLEs
z$s(%X90h50*#3dxw`gz#W)Z4)z^#B_-P>|VEKp2v82T9VN%>B-q`|kFvbd)|;?m!!
zoiD@i3=ltF1pCc1uXylw<$c8=O_;hf@k@P?d8p@&5m7s4>fypG4i#UhcvkQC#iJ+r
z^PcC%pn8E7F)O=iux8v;!wJL?i1mjlf<*_pG8R#|BtlL5a#y>)vCtsEu7#m?fBV>2
zHMoB~yz!S&){&aceYgF62qJz%9SCx77MR&Au6~y56FEj`K~{P9H~5eKeC5YX+{9$!
z9w=iHU@LlK#hWgjx}&(P33J!;F^I@7klLe#UHO~r_NgI=-VyTlO&0bOQ_^ch7zg7&
zcIL`9{#@TU>YBfdLwX$N$>?bz#e%{|x$FoYAOyES!H~Yx*+%$20VJwFF#rS<15&ue6nb8w^Mc*{4>0#hE~rWQRkG0g7Qd^#f0rOIrAvOc
zlTVQaq!uT|msnKG!$s^x_%9#Z&tghs{JUc2Wh>gsMl^u2{miOLr)cfUfI1HB8yGT3
zG_4{`q@o@v$D387lYW%EDL~tbR1&S9P3~+yAjfaSjg?M$$rd;ex56UThxT>LHfudD
zy$TUyoeLv;br2oO%n;at5vkLROQ)InTiZ)o^4QJPX~#5Sk+uJe73#{%8@lu?0ijDO
zx>|D1ib_-rwAx>M-3MweK%C?q`aDGJ6)_2lnFTTZM8f^-
zU(--(d~P)JwUHCEhFjBD~i9o<&wg_pHjH~^VW!*KoZ-_`uv
ztIZ!yate~P(hE$WxR4>Hn}*rK(K=A@G}om&N){hOM2=nunABp4M(60*B({(^e$q$h
z$~NCKHQo~=hQ=NmPIcEY1P{Y}+s^IgI7z!Z>t}$cn{vh75gHXT)3Fy+OhLsHHqC3@yrW_kPLr^z(0u60a
z8-EAC8Am>AGx)WX0!>h>Qy5ZbqUhGw7rt@p}e6?QBBy-Y1O*0ejSELG{F#RG|w{04h*37rs9yO3~
z;NX?!&I4=>v+}$_Hoq4~K8PCbROP)nO?;Muf0yIw0+u-wqG?t{Ufohv36jc@G+rSq
z`ZA3ZWVKa7IDuG620BfUwSr(^PTJX`SF3U4U=5_)rqU*DWI)_$ZBS
z|G)ytu@Wrc(Kr@56Hc~Q+Q1;KWOH+W>)y0ev6CT@R-OW~?-9v)7lf57NSy#w9Uoo8
z1jtnMr3URE+4t7t5JcNcm)M>ou~Xg6AcEUxVy-RWAf0N?|EP>1O(^JMfGWC)VcwCg
zlpob@9jdt=8UTU~3d)0ONgHF87={&V<-38mg_&{Wd(*m;r9H8xK?WtsCcH66G*%?U
zY4tTJZE+IONcqDIMCeC4_b*M3>OW_P%r_;Sm(So%&IAR74c-%mfGI(BnuDEVhgWll
z`+{e8S$iRvZN0S#8mvX3id^8iqU*5D8tJ+aJwL%rBw{HOX;AeylWR_nYQ12Gh)hY(
z1FjGh-
zOp<~D)GzEUMPJPzElz9OjGs#}-*b^T)aJ>x^c@`nt8i`GqbW7Ys2?dd)LXW|65vDJ
zza>#ESC!a`hXd<-o^6Xa4Lwu)c_P%y0h-toTP7naE
z7?jA6XB>=cP0j-q7zBC+uxfxxNtFR8isugZnjc{a(F;I@z
zV#DA7V6pr{<31|YeqYqx#npRsz19!T7>pQzC8^}j69;3%mU*ehBl)pbN+)CiRhv+H
zPJr;83_TmEeWn_af>r|XmFb%y9E;O6=a*CA8U|l%#pHy$<|3}#X<^Zxn(M!2#-0bT
zj$X+JyqW*uvy!8y5D)J##xdKJ69W}Pu2*{nRE%r;HN4Fv?x~ZbTKVi>9jA57
zfw-}VK%^yv3#%h@*|nWO3}lOLYlIOmlrSe+eG4Mn)Nk+0&EffM|$^j1VC@iCr~DcSg?t8
z%G|mMguobOx2OO*x_0$@kJ!#-Ud6dppB1Oo=Hb&E{voRm
zJuxoBn%RO3SKH5&GzJ!E&n6cd7>u5u6bsOODeCVt;FIM$Z^?gwSG;?Das
zeYK?B4Phnt7ux6n<>kn}#nm>zM5xq2BO^tg1JK}@Hj8s1+H2=&nr&CPVdF)TjvGM9
z2vvWkle0BSp0aB%UL@&;R-?$HOI}tPq+1`1i_;pKH3KE;IY+Kc%5sasX$
zP$97wcq-&VP}9X^#tsbf_83Q3D0=jaCyZhr+d%Z{$CMesj{Z2A(nEDD!yvDm3VFh=
zRnNplNI;>Y%qaQ_?W;J+FL%3Zn%!cL^qF~+wt=7s1SRsn{5d6JJ?;yl33@ud3VA_t
zT?gPTFAIPD?&Wbr*P6|w(OJ{r1U+#+0rt{mD|^@eRmWocDtmrB79Xc&(yjFCN+$q>
zgWtKzV9i-DL8Geuum>I65Nz@
za?n0ox`#asXn3CK^B!W1Mom?)v2goX3MZ)R1M|S;a;uP6y48SCG;gU44CTN1rdoQr~k^P1Hl
zdPak-{ut1=LcG2XBHIVI(CXK`UDt=IAtnQ@pwAWhLH`~IU@%Gkk3{n&5AMZj;Wv{K
zdz$++*6-ubmazcISam14bW&YiB%{=zV
zuB0vi;{dq5dugQAnH2Emge1=`&p6i^g6ORP)dW^69Xw17nP=&(Y4*xfX$kk|p+QTq
z12WRJS2(?uy6hG+61tn#gI4?_W2NdXMp)UfOV{H*Tp%3^skSbTRauUNz2w$Y@D_%;
z<>{ctsEit_D^y}52|88Sg)6e+$dSk=R_7FVem1y9I#SmkrFy}GO9aR`^+Bxrqi+^w
z&{WF?DG3&I0S0-iFcnY`x0>9tE*<~Q5<<&G7?z#+E!kE|^}J@5jCY
zqpsJXC~n&n0FnUi0Q$T}UoJfbaw#Nqdv0Uv24^T=&d?mkg}-rNLc$cv4j1qQZf-WC
zug!zzi#Fg+UglAx4VDkep!c={jmH*7c`iHh1f@yYqt%?E#M}KpSZv>kX}9J>dVq&+
z%bEKAG5f@a>&|+l*XF(TXKh6EKxJ8}D%M8h?0=}cyi57IitF3@U-y}9*2ZecQ=%0V
zxLvH6dfZ+%_OAP;czEfRV~7oXA!CL~@ZOHAMrdUY0fGfez&#I@eD7Zk-cZqCYKb>e
zmb2H0z1?z|z&=JXnbW83?onu#HBy&X0ZB({n!T|ZIV;siZNo}8n>*Wo2`NXy7fl`c
z0dZ%iiph`}z&)xbWXzu%PrN!94xV-NdZBar1mXp!y?x*_%Kd|~aSZZ8>6?rpGk-1r
z50e|b;!U4tM_RO}v@B#OIBav|)OsDM{P*Q*eUIV}uT3Uou_#h8>D|-&AnpGZ@JtCU
zcq+$IsLD5MJjoROG+!Ci872vgHOCGFZ-pHx730h_KWC6kd0tCOdK^_w??6io2ZS+dslp0QnwBL6T=@2v)xeOr)T
zeSNGCIA2zUl+y|X)0Shi(h{!g)r@NpaNNevDnk^{9TO4nj;d?bY|d}UL@AFoxO{)gh8T(D=CC)W9-|7w_YKdZASr&2m=Q(>R6y&IUG43QSd&S
zEjNDOj-HaD(^(&a6%!paFJ05o!t><>UZ$ZCPY_oXFUsJqLcg$G9BT3>F
z;6O~_Hr?@w2DHL>>=<`l=XD=VGk-c-Zh_y$SEM^0J6>%4=+%7iX7!A{JFh;HyFQCc
zLXWp7n$XRwA7q`=1F3p?jP)))It%||eW&5kQySo?LMk-cor9?=+9CF3Pcae
zRCnI{K)IxF=Vw1XhurSI7(85JeKTm@4w?P0%Z8=?RNnAf{D;FJDa=ojqMHoC;g+Oz
zdFtRb?H@ydCR|0{1bRO0r_L+cZVi*9xepVKUeYb`m*N4X$ngK
zg&5PvkLOajR)&6bb2r}B|BiQm4*SG$+9ic6?D(;{jJYEh{{E-TM8UT8?mJSfMbP#7
z%i}U#vC96$e{M73iEIvqKf%Nu_ExM_IR7s}A^RSOZcs^J%hkCAC?CTCQ=XYOmwe?j
z7J&QJ-_6~)^04;!3B=WfYw+1`1Fmem!#H?uDfp%&eBROn;f{CP)m+!$?>4SeegoWw
zfLGW4%VsseZ9tpWk)P*?QVmEHa8d8U?m0VF2O6PNm2ZT803TJQ$zj#GI2i9B0vZBE?>J=nfzOt4D0&+tSZ|=Yqfpn
zPZuxA4sDB6viN_!9Ac@XZD}yo$`oVmLr^aK_xX3(AGT|#;`;t`xz#rhDT}c)HgEf{
z;{VVPuamXd7+K$1sVW-CT?Uu`c$+T8E~p@MRbQiySZ;4pRSY+lbZ$UARaGJ@j;FOU
zQ`V8tQkuZMsOzzPj^SOydO#4UL^vtbx;+9Mm08x)yr_l{pYI!T;n8#%e2
zRV8?4?}A=CB=7TvQlyZ7IEg&FvycNk)-kE}Rume=vpwBgec#ih_Q}Qhs^aL@Tc@jc
z?OX({NYuthH8?70yR>mbxZy2;N6JDbz7(~NA*BTh)Yl_99XAnQ*DiKpuUV_C
zn6^CI`S*h1AD0)O@fNMqG)q%4G~@t2c`MKGux)Ng$cjrBes@vUNH?~E9p9PWv5Wr4
z%zJ#ro>zh?{`bZ;Wpxj(sp>X^Qz)DMm2NcapWKVr#oxX=Rhr?@W}p(y3fHu~))1$?
zs=V6wP%^_w?oq>&;3@p%-wd+8@~bG^@Lb5_3+GxN8Gu&sxv%U#4i=A_S^Pw$H$rR|
zhpLwDaT+%?4qp+PqN<4Ns5|bLlW7{CM5~^=55o&Dt5jyi{fv1g>s2aqyleyG_2C2L
zrl&ETg?IYulc~J7$NypZ45;eLEgJ#NpxOl`eRhCL<+8SL
z`%y`4z6l_;wlY!0f{CK9obyne^$R;PsOT9=>gn8~eRP|uggxsOu9=>$s_ggQe-Msb
zm`D0!hXUJvE-i;fK(6?zrcZ<|Ie9C~WeWbD=kEO<{z+9;+C4zby{f`^OTo@(1pFWS
zUI)ubXn4rS*7CXsSEohhk@S6ruL~X_bm&VilsT___f^5jfbhF2!g18lvW#@TY*mmD
z^LjDCdFA`BTaVx^IUBkqSZ68ud>(RTK6%gIinegHO+-;d^EL454klXh=#7J(dC5=r
zl4CCfWGS$3UKXE;ZxU{q#!FUva^E>!U8;gts&P@y4Se1@k=P)Rib8((WjuTZNzoeA
z_`9onQG#Lpi-XEKp4=rgydYBaR*Ky|?;tpq>_-)SZbKphj9*WzDjY3xa9Iu~bO`kG
zs7C(Cf_U(9D&RI&JOnzhPNf5v)~2Lg4P3rsVPNdw-#G6Mw@uEM;$KHaW#{%#&mLJd
zZ8jKhB&>{jo}9f0b?!5L%pxI5`>4KSLWij4Vp!v6>pu;nFHpnIQEld%UAFM=i&~+=
zlWyDwI{N#xj=JmTW_tg2x5$;>G0r>tQ!cLV^k(MRV|4QG@s}#6Z7iab3?HV__DStw
z!yVGhqEh4!r?FwX8&~-%Y8Dth(ng*0=zn$b|9$coiI!zY%Lb~XXnpYhn?LGfcT$3E
zgZxvHi|hJ5T?GzN9d&ATpj~%V+ST|YNqB6h){jNpzQh&S!#j5ge#2kBIBr-kzl#Z^lFCMkGKS9XRbfC@
zalFn_#*@DNV`G!C1oscN+w+suQP59cq|DZfV-JeLiza+~c9iQJwN5&MGJjq6Qb?gy
z80}Y~ZfAplE!|VuQhn-4PRDQhpG+^*;Zi$!GV;d_{4J*v#CG9$%B4{J+4!y3r;TGD
zUZW*3ClS)r5*Uv$Sh+J?_IFEEh&uCB|ZGFF=YT_TB9%z<>R(L5+lsw$Xuu
z?4Pz+@3b24#PXxvOhWJv7WmG(4BBJiJE9((SejV@tRN;+78w_dRjvZ()Xhp`>k%}+lGj5Uz7?(rJk?F
zwf~xH3ozGdFFo-nD1WqbAv2(+o{)Ccw#l%bFWVj3ApdLltHdvc_jKlC0Q$7jtR9=ytM
z8cp;6Y41wIn!3_3l}AAslez&_fm-lE6lHYSL&#GUsAABT%7_R;+1H>zf~J8Y)i9z&
zM8SyyT9hzC0NFwk2}>%744VjqEG#M@OM(a_kig9xZYn+Gw89k6NOvm^iBwPDdeP>BuD9&kx$~r3))6&5*2WAz2
zm)Iv^Y4o-_WsL|$aEOl!2{Y_?XUnIH1TAK$f5F9-b{~5qdyTqEMR5WuMMo?!QHLu%
zZ&op(nHNcYll{z@dNSYd?YC^x>I`^!yrr{Za~RP$Btd9DZ{gZA<((S~KEe1Z>SzQe
zLR{DVsq4Cxl4u&z&@EYa5qjIb|1_Imew*$t^aB^FdBJPvX$5^0YxM~B%m6rf&R!U4
z8lfEs?(EJfYgdOc@M{w6SDUrr^r{V~%VPs04G$*NM^;k!{eBVl
z1?~Nt5yIWhH(@1Z#2-j6S$YuH4yErUU%^kmNODP%WTOn9CcL_E<0Kw=sgR}=*Y78q
z|2*h?;HJtQei%o*){e8VR-*YvVmcN(nnTNrm<;PH8bRg6(KZdkFi>bu;E{q4HmGZ%
zDoCYBY2ElNj%`bo#;?OQ3W`@3JNgmGkv8rQYNo~=={c{ae7v{l5qbv|&!iuVpXPqA2nUhacXEm*F(B?C0=gwfR1@i6?WHme)i_bRnz60u~P
zTMW5#zy0qCbt_PM9OS~$+bB>!OrZ2jMs+b3%7JX-m=n!(td|VE*m28WIF2WV?{Yi=QJ5v&O3AP+Do#69=n`k)sz9oH
zxulTwsXb|S;!AD*ip;rlD;V{o)vbpoErUN=Ov?quXh-(EtUHgze^xVOKc?@@-X*wO
z`s|aB@L9^iy5x57!X~C{B))T|saF{D>mm23eROCWdzyu!U~6&3{VM^u3<98<;JJ7H
zK7p%coD*L4WZTH1aviWK#{)qK#xEoUr>{gepfzU^C+_qmlKUm^X6~B_vi!emKX;$h
zFR!1}Bj~hs9-8vgS1d>?Xs-C3?@xh;sjTFA|KAFB(WRCii)n
zkk5#hH>OI?AdOVGO%Kc(;6X-V&^RB$Wzw577&%$P6B;Smz7xpz&
zW{Ez^jiFfM#0{vmX(n(sxu;iKU6g7v^A-lgqUPo5;@vB?veNFg@`WYd@>W=y8GF%L
zhu(sAWh3{B{M`rq-hM~U=k$6cVAdU~lG%kaelD(=v-#@{ZZbh90#|)1KJHMhY1h-P#)t8BnD@-&k+=te7qx#J#j+nI^I?``!q`zw&Vo|0z`cNuscV{@(vikXs^M()KJ1oq}rKbc}!b
z(l%3UA24e`^(bLO+lb_5jK_qWR4J__i(yADC;w6EmzI-TZ1Q13c5CwC5M`Rgoub?I
zfYv6(`cj_&+)3hI2KNs5Ht@1XqNWDR78oyinDcW};@&h6N1!b=6eU#`0_?^ggwC-h
zj-NtZwVHU*u8qKMc7*|UbaIQHnF_*Ba_8L|yqLJ1+I9fF=}y|n%;ly6OCI74m>n)p
z(BDZ3)1IZKGOdfpGX
z=LTgc+iv~F2OfbddQsGms7)krtTDyNYuhiH{9BSf_MDvG-4gZm#8iFgn@e9Kuo)^E
z_++JfS(Z%LU!mJcN7v;5KgXvOvbQ_X^G;u9V
zu1`WKT8$bZnRSuvhX`g3Ho%U}jOG
zN1|W2dGN=v9~+iQ72cTxf*ZT?7M5gEuiZKmR&!z%3ag~B%7v>{;eXdhxukAs#2Pc^sy)X`+V@}W<@XiR
zvmDZga!$1yMW6Q@z9jT7rB33skhQ+4{L6V7-VCL*8@x|8xxH}q^QRRoffUuAyEtAy
zL5fK=K!GQHy7a)4`kh*JI+fbN6jr#B%Q$PlFvrc6Zr1+1i}<90xK|JZY(>cDJGlg%
zMLR|lBI0{?uf;lGj7!BD&}BpFB8Mow%#kman@}+G!(2vUg)*4=c{}YI88&OOg|skS
z9-NpUAA8e8F<5#%#Kw)%9YT(MK0ICG1abgh2dPmMrR&jnV0TUa0bn=kx2?bI3)6+N
ztT9kHVO-gor2Vw)&;R+nuE!2;dEOU_0mYYL8UnmYDmah>_5cbE94k-*t
za$0bE?f5Np%yQGx#=E#$#rsi;0cmkQgLXU@O@snDa4)?zQaVutY+Ffh9~{tkgNulg@T^wkgSk_zT#`
z$ixrv&zvO>ktZZ3Hl%KyIG>YBK_RElzW2WG-F;8*zBU>b$n|2^nV<^5uOeA3eHmOG
zB7jdYhB0E`BL|@(uCR))IhwCKy2>?H^$o{RvCgn>I;M)Ntma#erDB8a_;sg_paP^p
z5ZV1AJ%7lrTv6pEe2{HOO(T2%)$}b|eT!yp(X1_+u~{4AA@t1OayHOI`Ht(pB|M^G
z;JUp^ZVVVDL7dlz-iQ(YJc$!dWSKQ2oQ@cs5iT=4U`Z6K@XpXXC*wrcF1bPIpM|of
zb|bv%-O8(@voMLdCQb8(r#$R(^WU=2U<{`=X-2lFV!oeXybDB#XQ@yGL{6gM;66rg~cCI9fV
zp5ktXy8;w&x8!^C*C}pixGg{dw@dzH{y4?047UU*;FjXW?UQc@DYi3g3sAszdEowh
N{7*w=XZ%3U>n}_h4ub#y
literal 0
HcmV?d00001
diff --git a/scaler/__pycache__/about.cpython-311.pyc b/scaler/__pycache__/about.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..f6f1b668b2ac1aa86a0bd16018af65d80173bcb1
GIT binary patch
literal 188
zcmZ3^%ge<81fG0R>Eb~8F^B^Lj8MjBkdo;PDGX5zDU87knoLz}hI$ryhQ@xHjJLSs
z2ypPQFN08K0J#7oVP=mtIwx
zuV0*;n3GxrqY{(yOH1?$Dt~d<p+&IiZ*jNPJ*sWMsU-C)U6Xf<-Jq
FApqCuFcAO%
literal 0
HcmV?d00001
diff --git a/scaler/about.py b/scaler/about.py
new file mode 100644
index 0000000..880e579
--- /dev/null
+++ b/scaler/about.py
@@ -0,0 +1 @@
+__version__ = "1.8.13"
diff --git a/scaler/client/__init__.py b/scaler/client/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/scaler/client/__pycache__/__init__.cpython-311.pyc b/scaler/client/__pycache__/__init__.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..c5c6bef5a31bed2b18de2fde2f8a8d7b993e380c
GIT binary patch
literal 177
zcmZ3^%ge<81fG0R=^*+sh=2h`DC095kTIPhg&~+hlhJP_LlF~@{~09t%SAsUKQ~oB
zJtwt1vsgbTGcUb7F*Ck0Gd?XfFFrj#FTJWXU%xmxF(lIY~;;_lhPbtkwwJTx;8Va(um>)=dU}j`w{J;PsikN|70MQ03g#Z8m
literal 0
HcmV?d00001
diff --git a/scaler/client/__pycache__/client.cpython-311.pyc b/scaler/client/__pycache__/client.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..af5f5a9831645b7d8203ebc5db1d683441eef7ed
GIT binary patch
literal 32127
zcmeHwdvqMvdEe|4v%mtoc)y7S2oeAZ;zf`WA(A2mK1B(XL{K7Eq?b#~f?RMH3(YP>
z0a@6LtJILo*pRJMkV7|Uj_p7Z9m6Soik{=NY;31&<2JMCxY$e&&%rq@+orETm6PiD
z^tAnbcOE;l3(8WG{@209ojZ5#d+z<-_wMhMmKJlkt_&_+`^y(O?%&aaez~;3N2@lD
zdzTY9!4~6!>fIK!v1fbG&hCz&gWa7$C%e0XF5K-gcf2TAWTQNe7$5fpJuK{u6~{}0
zC2?=i8!ru(#>;|babM6E_Xqv)@?d$qB3QxF*LqwJcl|
ztBcnM>sgqOHN+c(jV$bmHN~5Q%`99T+ZJyLwy|1gPrlNU{`!sa94bHa5szh#k%8rf_qrlAM1(t276h!Jk}TQ5B9Tg
zMQk9xH@KIDD`SK4eZhV4{lWe0xhnQV{K?>xEL-
z#9@8-?c7ODsCknUYQJOWxDW8JUf;p(1O0&uKF!kAAzi&SUGRutJI8f3{28Sk>9Q$B
z!-?t1R76D3lZ;-Q45y~WNEfd-j!w=f{BSrH3tx>z6z8$%SW4kfBc=NMTr`R2vs0;P
zVlo_4T%*%dNai{}iD#5mG8~IWCR0bR!NX&v`gmfRN`F2)8NSA9^qw0F$0DK{jbz@l
zS0^H4sb{9g$Ju+I`utJ^kCDl-2$GZ^i;Rb-W2tiyF&d6VXIZ*p)0178%<0lVu@=}jLNx$sD72w7sSMPG!{YW
z=ysP~@lcT`V&Q8^rCE%m#AxL8NGOH4&^Seg#>GUOJ)=YqGLJ=)N!Ed)a|q${D5XZX
zMrWoXO3~=p)KNhYmEzNrQ%FzYF=H+v;LsjE8baV*%tu=1LHl_w;)pl}+d;%eToJcm
zAGgC3yf!)YouzTL8KH>E$Vra!rQ?1!hNrju})ic@mMDam_kqq6p$1Dl-P6SJxQ{P4rd=EbpZ
z@6A+p$dw()UK+DKtZ9(~6U*n|sV-A4W6SHMSTGvGR}@S)ITI35Xw
zl;Th*o)D&E6!wNfuT6(z>Kl<(n&JzE!jqGURG8*QG87U^5G(eQw-?^~oH$6~0rEO@
zFMvSuB)nUuzm4LGTO+qdesv^U>|1awoLk^;_e#!sxQw%Y)9GQUybKV
zJlC-2sz`05E>bU4o55uca3Xy}q(1jN?gqkl1{*z`Fd)=Lnj-E(ORXU8IMVZ26*vG2@Qh9K;&kgWmVz3-)0x^&s3Y^kU`=2z~JPVpK{o
z-*{6FyU<_Ir$NCFggB_URl@9qQWOR)1FA>yfUp4-1)QPyQqg!MF`Wt}BOv62q|$Ib
z5*AZeBVj@dAi7=;$BcKyNrTQDBXFr-L_RPgoYwzNI%NITQ6?|?k{
zz#vgT>=PMo57z*<6$i9y6gc}F!(Rcl1A*ye1ZXx9NJa?Vg2rQ57D#yPdW1l_*i&re
z9-0b^;dnsn@lb%sICPzefpdCtR@U^@cr+EbnGkQ_F_%3B%$&;&B%FJtmTU~+P~h}<
zU{surbTd>BB6M1uL_^0Jd7lj2jK*St>*3cyfuKiM7;P#{iwwhbc3laeps9$Mnh8WE
ziJ1_3ay79=m@2xONW^k))pq24zoEc6^+N{0$V5@G{xt)X5dEJ{#Y;74X?Y>k9$Y>j~i8KljxT*<6*BmEWyo7*w+zD+3$F+xRUbV8x`
z-C`f6mq=u_c!)gW3&p3&I|46BGgO^R3||uwpbyqvmhPEsLq7&E>xWJHe+Lh))h*B?1doE;9=Vp3#
zg7UD-3Y0YidyDjwQYz5hASV&|0JJRCO->1oe_#
zsT}O#CQ9{K!7ezg)Vq~RN4aHsT1&Y)X~U&$1v&hB4uB3WRBk*u@l>HdrHES6u8Ar=
zRM0Da5k@ruy{|N1T!IP>{nIkJLUR5#=z(5n-e62AuL53%wC`qN$BI-qkhc!XaLRqGhzw5JPIEbz`;gSGfLGMWA#FiEKR0DI@&Kc
z5Ty}Sl+>gkWh%M~#Z_^XQk+aBra};=Os4YSYL`=_fdhjI6d6*%?e
z;s&xInTaQ)_gAFRS3I!<$c5`*Xce~zjyw@pis%kJqEv^r91)(0hk)IcicsO;_(OTP
zQE3c49$_h-)O9fuMs?Q|*Kry+>St~gX^sO2X#+0Of&sf}9d?mGsp3|@o8p2zKun1>
zh!eNd3+D^rC`1JzV=R>-2*|Ecjn3%gctWI=13Z~vg2T{NCOK3*Ox6L?A*mDrou`06
zMFP0F+Q-8r)tgaj)EsEzq+tF;m1BuGs3bLBEj%e6LlOBnR-{EeTdht34ZZ64n&wFI
zPk;r1Wc^&z&UvTI*KIUxpLgE&J#5-`=k$AL?wpb8$^D?}Uc>jB?lsB%!|VMg*ZNOp
z`cKRKr`4D1^)Ic}zm%yTmFq{5x9ri5Zh6Q4AH06=^Y
z_OPu>ZW~y6BGdMi-1ZbjwPXXmbZ^e?K2G<}M~&@wp3XFO%Z=UiT=%f9Wy$tl@ttB6
zU;3zJU}aF=JDh1bCbt};_e~FLTC>&l?|d%XxP*7^E1zCOb@%6y~5H$H0U
z$Wn3NZI-*9x<4u(4@nb~^28euZ8V|)ct0nP2JSC-&f~xBm8zdzsgnCo;SQJK&&d24
zi9hpb+pguoyN6c?<)_X(82xD|v+c6H?K1VD@nK7++|u`hrUz9&ZIN0oW?C-FEf-nu
zHh3@Iuk!X
zM@77ek^u1ga8vlm8)G2?`brUnz!(A)Q`FTY)k&nbL1;(bU=l_BQ=-W%;)_Tp&XPxB
zG=OQwELjBJMCh^)sq>P#$%BbxOf!V3bL>Kxa3YrzcE-A6-hBxEf7r-13Eq(3*5XOnGA&WQbb)JqookQ@ujYV;|T}E
zNIa1O-v?H8s5r-)_VfTb3X!3qz|>6YdSWv7hQ?59+#!wY?d<94>9VE}BIA&6Uxlh^
zY%JQTrXSLhcLyeiAazc44H=!MLg|W*2POlD4h8xOU(svzTLk(F@Fm`Ib6oBF&^C#_MPMhdWs60EKy1l>822vbdH^N?tTh^PuAm~Mjf
z4Il?nW}(CT&aF6VzyM-rMJT{I@d|li^76qXBhwL$)Llyr2Sx
zxVk2}wkKQPe5WPbvR!T&%UggzhR9acFLudQ?b({f#WQkEXSS(TYTGL}4Q4yK
zr9G$R4vZJ)Y5I_Jc|aGy!^oLEb@=S#la`?)FiS+>M@IlsA3t>Y_|go5m&~66MX6Ul
z2cHAYbIug6r{cIJL!yz31INWggjSwy4ziRbLY4&=i4q0dUAr~?CA4eKJx83Rfi6Cc
zbi_uKiqXO)gR?YOn|50{f8;5`lyHPkpekcgo>PU2;Q(}cP|z_15a`rZjtoBpbd6%N
zi#R00N(KXh6%PTY4ze^sVh+M+qRnvtdZS2H+ZcHxGB%CXtRbG*jM1VYCT*iE!vnv(~R<`F*JtJTexs?}?V=le*P{5m{<%Niz`>|EnJ
zGyE=@-zD+8Hkw=I=Dzjj<7>^w?_bR{pO(p;llX>*mG$eD18bE7D+4R3)t57s$K=Xm
z^ZZ6_X=lr-%_*-avP_?&x>wME1-?TWI;oHDb
zNqk$@S3O^<0u~-Rc4h_BGe)3TzYH?&e_>&KiTe`&E@!q~5NvO9Y5TD}xubP$C4DJV
zi9&(Z!du+d5GWsJy=5W~mjKWUuL785z@{(rDMPx+ar68<2ZfB|O}?=0U*QcnU|^G9>eJ^A!@E@enVIj)@(2j+@W
zhBOayFT?IaaHWg3lDA=$O(T%&N_7_T3os3fsK{|SG$I0@DY
z6{l8GzcI8ZeEa-&k1ai)sT`0i2h@18&sNCnF@1#ulGhW{F@fwts9gjk4J^>9#)zt!
z%AQ#CMx=`p=rF}HhBT=v@&C~gc)wwPm%9ekb-7{A_LgG_F@@J#c2G9lU7HnAzHPtV
zNW9Vs@$VoDxbnB@<`ln)8yhb19STv;H4g9AEe5Q9QGAe^3#ri5;n{6lj-B!TUr}!Q
zA|GSqw%yQN@N?NqJDSWguPV6u5)h)YQJ%ewV5q)7(gp;t9esBYP@kbHOc
zYqM{s-$@&h>Kxb=AuAuK0jH)g=fENommolusVX?fn7FnF({HG_s|=}%?YF8(c_@v8YLR#-p0vDkp2n02JB>r8B
zEK&z1sn{ay(SJ$}&AaArZjHpM
z>ZyLRAo6s~AIp0Ex6iJ718d&EQdP#gQ}*tZnA>!^JXKltIhSwZ3ubDbkZYcpKRJK$
z4+IIHAQFw3}?+dc`16GIZx;#JLGg>x}~p2XT(l@h+!+ArkvB
zD43;AJFFbnCFCYjevSueZXo5YnE~sHXT-dLe79D5Zb8hO6n0A_axA#eej4|JYiosQ
z+HFyc8EQ@e9O*R$HwpKEh8IEskpuvd9Z?pAz&dUgx_FPUzz?5#S>=oSU(ya&>d@)0jnky;4Zp@cb5HOCp($ANg
z2pBK$V^R1|K#Umti6*_Ky#tNB-mmCfGU5@UBFx_*!82HM(`clwhf@LI#-v(5YaJSU^
z6v^=PY#LgI#89Mn)J45}XclYWA!;1SoRd?LvFLa-BA6;s`D-@d=S)m?uB(hJdEccJ
z(N}{aRdFX1)8be};p351SO}-W3RJ?AVrPk`QENO
zyNp+#g`z`^py+B;IKm0dG(;4P7!=2;o?@(SO}i{kAwq1R*dj=hL-FJ_@gGpsMT&Amkuwe5x{HkElj85t
z3!)*#W%3yHSf)~szef?@BX0$sQauSJhH6)>3Qw2;J$7-B%!4gErI_8=d{BJalGGZN
zRQ%(xzDmM$PmdAbrLupQ%64EOh#%17U3kedcsjIH?;0_R@imB1gdbAk5KBqZMs
zxNJ+$a#C*DyHK>~c-y<+-Dv4uaX^l5Jo`W}+dYjNl@0SF8g)$zY^C-eHvfL}_gn6@
z;2~3URIWMt|CY-BN$IxZ-P}*Qi%)bqezJehiS3R*+U`PFaX>z(IB5%jMSMlHAAysP
zHsF1F+S(#=Hc2p0h<0s3aXT#AZydSpH)%(X<~569NH5ThYl`^?Kno2j2kVWInah7O-suPg!r|aN?_O+O~@(9n8cj5+uUHIc$Exc
z#euTOMXPiLef^=pwJ1zML)Q4%_u*M`wh66eZ*
z$uCTd&oWaTV^5#)x~n+H`zsY@(Y9KmNc6N)qCPMntO*B_({b^;^v)5BOe$V20W+V{
z6#3%6#A}h{GK!0Rygc=N35}zM_h&0KPTv%_lhEa>$POUePzYo<{8h4l`?`Psnt%UF
za`jloe^mA#op(L-ZCjj%&B@A+d%Zu|`;+9q+$+6sUVh8=aX+@GqUcHtOQ+Qb^2eIaHe62N|YKv}Ok-7CKz>GJyBDdx>KAS-<
z<5GW#$k$>hYw(_Cv2IR?n={4DTk#m74~$V_T=-zD3ml1dQPVx>bj}!gfo&EMUKpFQ%O{HrV)#w?
zZ^itHN4$@0x6ZBz^8V*>hs*FU$ovZu|AI~zSV#y&>e`r}HSdv|_x)`D33>l{dHrdzFd~!>tN3&>8@~9Inipis(y$7ZD9W-IlL98Bzv$Lj9Iiq
zr4(G&`Sj__{$5YomRomF1R3+5wzFF2HmMy*iaj#hbB@3@Z3H2<2xta|r73W1n4Q@3
z7wLiSbFxF7kTf5=+^VL4q;RT&5Ax5k2+#o3c4H(y1xHi-GEv8DkEW*ASBkKYA_>b{
zv&1Eh$cLhj|g=O^lY0pWj-OzX_`YscEgry@X7h2hZRkWul#K1!A!*=x#AE}
zYJJ(7hIbO{HG9@-_GD^$<(l64lYimhutPF09jG>ct?X-|OlKi-ZIYUw&s2=a6(f>&
z1e+ugslF(7=P3)0Ryj9=&!w6=0}i8p{xywjlDzwS?M{
zm>iE@gPj3rvt-^X&q>43ibp+)JyoL_|yZa>24B&?Nr+C~78;J_N*aH2C|CSYSPV2kL+TpR#Bp)j&yxyo;-P}@-
zc#(k0Pf-I9QyJ&4y8SEbzU^zi?HS(=*|+1D;-yhC9DYfv9)-*B=Vku9#Gn6cc%z(5v^nQB4*Cr{B;mrG
z+e{Ot-DXoJD_#^E)2_!DH@S1vp%sTggFW!rT-*fpf?c0ZkMRk$L;ySt9#Joq?$$Bb
zT329AGgrinI9vs02mICs2G9qYwu(WYU`zA)8jv|?EVc%jX;0b#4YtF|yFyRFO&Hwp
z)_Re2asF7Ntm1JyXa?tqMMUYwI4}|A6$CUD^KL8AlAAOsq|vWjY;N-8?B1X_!{p1I
zP6K6hx2B8+N5;f*{88=r3B{|27%|GL?BVHSpgovY{;R0cOn0cd>g@)7p&AjHgq(&R
zUL;ts2iYOytV(F^l`Fq7`p)HV4}R;AydCbu15c*$Jhl#qrm8KCX-YJmg5xn
zhw!>eRQV&bMljYn6JaXF8iN}Zl43vA=4S$J_3#&U(~!q7vS0#hCS^<#L1~gGs^UcK
z5Y?=(bV#a4jML~u*^1x32kHm$rXvPsZOT`}-v3HjEeTa@)T3wo_|uryf*h+Mbum
zy^`@?ko^}V<~ABz)*JiQ8v8Ph19IcQdgH;h#)Fy0!*b(c(D)nbXB8&#^)ginxhgU5
zd58@Y>1tcdk)aWsVX!w1AG`%j
zzAQFAoONC17SX36ZwjD}L$E{7=qQMHTHtmb?_hbCw+-|L5yPAa-NZT0oDSw+TNqne
z`!eS;Z{E$%-4dyEk!rGCJfKtK54898bYHb6^N^9TqOjpycUo+W31II1Q
z6@g@RZ*3}zmKbshb92%~CYdAF#|QXVul&~B?k_M#JBRI@#EEMc(-eQEHe1KbTH_pV
z*2$+iY^k>5aNe38?~Bqrq&j&fc3;eyvwdRQ1TJqp_rV1DuQ8e!*;0LEtO&%~ZqN?<
z&7+ua)c`!lTk#w%N`~hu5YX_PlT4|AnQ$_ds$M&QckioMN3l&6;+(w(&e__Kb}Jn8
zs{}=f-3AV-9#-Ar(S@+_ur)`@+Ile8)_Q7dPW*?kG3gLCc%$p~*SYW2vR*al_|^e)
zn&_mU_+B`B?)2#Cvm-d=?Ci*M7nN#r#7m(w$1lEe_N8Ou|3s~SGkyPC)c0Ps@8wq|
z<`!+s{df1TgysDwAC$|duSomC(r$7Kt?C_q;gC2{2pf`~+>lf}en_Ud@A=svRO*9Z
zC;Mu4Zk*zZ#HUg-e}mner*7_4yV?4j#N4uN<+Xb=_mBSgr1ZoYxTRM04*wH+cb)Z)
z>4|$E-@6;^bAPYG{NWL%DbB7NEcCrL#L>91%bXS(rZ>uj$Dx7E?t%{MWpD2
z9GzObG$mb{T6pc9*B4*AGlQV&3P+GeWH0_dYKLc7*p^twP_vu$8g{eTjk$FP{~k)2
z9r*OCT{o)z8ugEwpjpkfT#{NYE%z_kzU4+B&#_+2dcgbRP(Ii}%z6#23rs#B0j7_q
z6z4Wdi9be!_^-*kL$91;2^?tVo}PmCDI)%NigHXPrc^Zte#axDT90H}pOIUi
z0sPEV56jiV^Tit~U|3u_hid!-);O_M
z2TfeR?hmZ_14}i()9~93%SZ2?$h7X2TlZ%CgR*~6V(#OIm5mTpAi3Fp*nn-zG>cZ;
zKS-_~`2OebeO@{hl%5F6eXk-m({f3M-*8E8xU^7&Gvk&9qyWy0JNE;}%E|A0?|Idy
z2Oa7?Q!^^pj7t8|O&3)}CYn4~QD=0bu2ZV(l*_wFZi=43n?IWMR6lVC%ct=0;3NwmgAXmf{v3YfZLJ0_LosB)XSPf
z*=g4}
zI}nTYYz^nz;LK2vYpxNIVECV52GC~0PTW)wiAPjM0s5r0C_jQQ@N5Ces~Dbzt8
zsfY6zaVBaI|1-V#FXa7KcwI#%)1H4uv3~?l@ffB(u#JG~gq6g$nrpC4g={rokGC*l
zGPYA}{rVf|Me;9E4WyE8uBL9j_+cWMNleR$X;|NEw6y=$uYBWI-kZBKx5*WIrm^bD
zf9vf0>4g;W)$QMQWqdnj-%bJt_21pQoO=Jj-2*GH{qCXVLw|7LzVM*`$Ja84M&v^y
z%ZDCz_G9ti;Y-f(AR!L#7mpko|**PEVMYkKDX;De~t^h~Ddf=uq}g6C1+;Jwm65Ple!4_}ZE
zho!MtrY|n{#ic}2>P<;766(r$yCrY;hOd2TLh|)Vd>=z~il2?J$!XNOX@^*0;d?Pv3w7RKgykNy6m~+sXM5cy){>kL_ex7%
zocad-YO7;gx0%6A9(Z#BywOT1xUA_=|1bk@wCS@br=dsuc36(Xb{y?##+}+%Z)ul{
zI6hboyBYUa1fTzAk#+RGXuI8R=KxF$P?>gLvzhDpZQD!SS2*xt$I`B?xo2y8wSB}M
zD_2JFv=aY}Y}2^1Qo}$J@KJ@h-z#Cnd>P<`P4r`CFm_EdgcSc61;DCaMqC21!jv$e@km5
z&@t=iK(;#4>5?7?tP9Mm1$xhNVMAs)4FfF|6)5ub*i4T^*kLz~AmHR1F#>un0+J7h
zq;qVwF&ivw
zabb=4%Kgv7gUeK0kSi`o-V1=BRoIzg_Z%Ve2tKpVTihl2+ToTfrNLuT7rBQu+oYDm
zOJndQ!2id$c
zLvWWJ(F=dUk*%zew)IGrZymtR?9f4v?{T&aN{xZ2v<p{)7miqJkH9t$pO
zI>5w#$AAhy=Xt9(Z8}4g=~4_C-_}~9v}b}wlU-IQIadN*nHTQ~Yq1M8eQB;VUHTTp
zFiGsq0Q7|@PE3d~B4;=6sK9WxA}Oq?uH!-a^=^6dX}*=@$`b?806e^O1h;bjbvv
z0rj018MR{gm89`gmtCP*mBF7x;ZMM4pF
zY2|$Z0gOd{yXI=rHA}h1>xJQ-HkEBY8ZvKGEi)97`7yLh-+$K&CC!^_YZLEpk;RM(fTYPwO@Z6OCfX&*&&QCSL-L(x24a=;f8JKN$UC
z@E=|J;U)RdS>pwxQs^9gGP(r=1rz02?+B;~9C(q8;@DlLhHPV0{5N!=8&EHv|g(enYNFbC^ANEC{-3*rR}@1Zmd?xv`;e(hFL3IPj
zpHr_k$z!zFPbd^8?=>({xP1t`?#^FSz;`h9FymH;H`O*z_
zb%bxMho{!UQ<<zc4?tv8Q+bv_?aglwv;#JW!Ce4T}>}Z71CFu(bEI)UCVP
z(;JmFpk*uD;j&fD>s9S*RqYS!n(mZ-cku2ZdG~R7_p{QA5otUszc?Y)-N@A4kn3(>
zPfm3W_T*I8fK=5zNE+&;rU9v%T(<4W)k?YT(A#GgPA!Iwl9w)Is(R$A9*Ma}4ed*x
z%QOti4TDnE;G@p&yWVVF3pM}q+4`<*%O1JqiR`uxDAT*jpiJ+=&Kr0zQd)c^7
zcR;J&_^QoVlnC5+!pk%S<%ZzA?_u>exq9c)+=_5-@_yoHue>T!O#!F3+Q5t5zQKLl^Vx8%t9+a-TH$vdm1x*n;VT(n
zyl80h2x@^5a?AGhmcwf;hchili2z>T2Q^ic|bW=>eYbZl;PQU`Upx-!SRNF~mI%8Ce
zX`nteseQuc0~}Rbx5n3H_y!rfBXy@X9f^#~_}ue|`gHvE91NRp*#%hJfO@B{7hG0?
zlh)@g#!s3hdil0CIjt|&qkZxZ2x=DNS8}&Maz|HT3Tb%B3si&VTxCeq6I4kkW%ae@&;*I
zuD>x7iw6m<(@^M#gf*y3c`Fbw+iF?VTP4{>-|{&W2O5Frnz6g|sZ7%$x#jY2f5*bg?bgRI3xdgRF?>4QjojOZcnyX28nL?l6j_*3#ofT5~K{tboRAn!kr
z_aDh4@T3x~ZzH65>D0;S81}?pPYB`yz5LJQb&&T@$h%J-BVYe1g&4eM)bX#=
z>+U@bs)@xta6UNqQx6N~dp+EwO~kK_owT{IJDPl)3BI^}aeC?8@{SeP%Jut>6{v!a
z-5-8qGTvcj?n*WmB?;6LvAb&Bb-qD%8clW5_`%kO)E&O($3*u$NYjeA3Z;;9N
zSGt-vIm6pk>uOpC-HzL;Yjy1Y-qq_r^sz_jrI*$Ft5>9OM7oxg!b$nn6pPFE1~_-c
zroGM;Sh@iA!4+6Jzq!93zNNPDGHyqw{(A(!;
zkX{^>UJgkwhUDk2sBzb0QhZ9j{u+DQmkuW~WExJ|C=+$>^oJO>-yy|<&AATN*vr%qLEmaw3gEB4LLEViS$
z&q*zPazp>h;HqQwlJw+hdGE85?^)S%X0wzf4Bs&D*>zMYAwkYqj3~0CI7uM&26-=$
zca^+RcuL7wA{K)_ihk0Msfm;#V0)aJl2Xlz+z<+Z1W@sYo3b&K!4L{VBMfCQ)WXon
zt5hgT>Ur_}SRp&sDIWvM?G8T%Std)8LmV=@j%HZCj;Cl47xrJcd?D?61^oMNZ>4t-w3=
z7{}Ws87Z%{=DlZfpPmGMSvDS>WhWGgM8ql`Tc!htp#RcrNK^cg`cT=Kms&p>rw2`^
zU!m0mQ>sWx|I&IBjl_&KV;d4w>L6flh`RGkn?45RM-`~lr%Fuvw1!}`X(ww;Y(p91
zcJkWE+d)NojK!dmCyc~6K(7t5;}`rv_17;Cix)wZz;Vgn096P+#AeHK6_W0xcE^F-?dYa{qOIDZV
zj!0IQWOW|Q;?tKp2>Ev`lrLu~^YRyvxkp4I
za%mFB)RyLEY;oI+J#Oc)Y)?C8oN?!jEAE+PSaSzQq(+x9?@x~c%+&klo
z`)Hjj-8AEm`)S^t4$L&in>k`5$BD?_BBDpKJ+Kq<75vRB-Xe07B--#FSR9Yq4AWiyF%lrwyjbD895
zHj|N(D#i6pT)qnVbJ7R1lA;=p^9kjBqXFS)B9oM`kLTEjQbs+MQdA^xM$S&9($YCe
znN6$7()eiNDHUEm;RyLNm@|pQIg!L|A}85Jo8%Dfl2dYRW_?75!qEBiP>tVRWX256$jMt&(5?cVf1+atgv|!cK-BQz(gU&+e
zmMg}IHGP;7lYTqQ@@8?f)B!y@pvRU{
z?=6t;6gzK`c-JENh>LFpUfX7PCzFY^B$qhNp;FFBDS{}a6LXSi_@?FT>@@-Fg_LM`
zl4Z6^mSqQ55qTK@FE0aE{EE9rR7lI0b**z%EoBT+!w(!$tEexKt4u~7K<}>@g#{w8
zmTIh=qe8io0}MvhN{1s0+*MZ@6R5gb-UcO}mF!R|O8{ALK#A%WY!?ZX8q1h!WogJ&
zU%A#g6M)}TZV@>s`7NaYDW+@QbJ5+rVS_6EAZ4364uWN5-J-bRH>wjNl>shP^nEysSkk%{lYb>kANP$
z=~OoAG8~GOo|18jWSkkf1wjyB-y;agbV5-CLBSOb<8|Zvk>M-Z8EJSrEnQD3!)c%d
zlO)WggefT_OlLFG^RwAug^EHsNv2aUoZ%^|gJaj`vEJ2bxIckSJ|$QL{F?Ys2$U5k#}z5?&P?OWla%Y1YxqVE{V
z^P@UHs_~<%+j>4*pcvhQiw>P{FAN{TbW0(+2de$6d-h>*>+_zWFa99kvsdrgi{)@3
z62-K;5bnjaYxMx_)~WG$zl6b&B$FH{V4lh>&f-1bv8#}p*Cct!jYm0o3ltx6avvlL
zqKD-HfHGK7G9R)`M2Ua4A|{H&4YKMF-u@UKEJvYexMT@TS;jH|Tpl`rUqHq>fHz@u
zFd#PE8^LOYfWq<{Fwvjetz!JAP+#ejw1L@%-vb+{4EUQDMPt6^E9>=$Od7e?L-Pi5usydyj(sbGYu9Ed5jymBu*z-V|c
zk)Abev^)eARnulPF-nw+=eJm9(Au~i``!a^gA{ztw}m2ccm^ma_=CE?NAnLpZw+hR
zXP$@v%sbzDR&PD41HGr8ykfK}4K=y%zWc7GrlzPM$1is8v;V#0
zp!*-WgYI#|Q<4ADn@S}~PBJA8o!F}X6jrz1s^{3M12(%JK(692$ZfU!P%#m|A)ni7
z`9efB&PK7o-E=RY!ewS+!3H&m!wdv)5{{H!a%h3%m`NztVp=mQ#AC){eLTwB%#7ri
z3E8MeeO^YM-H@BVZ%vuLW7yPVVFi82gJFXoduRNe$rHzpMjPc}c(CCRvl&Uo88KV}
z#gex{fwHySiPf8gD+#5noYUYbO3*C}pbDI?Q)j1f#YY^@(BwCSjrx@C+6S01%BopZn$9r$4*>vm)VqXSnCB{o26z
zlhCs*kiea9ozPn+w7|q_M`ZEn?RQuG?V3LVdi9P&5A?12yEK2a(ElcuB1IPowd=ti
zq!i1kEB!~7`;X-NNA>=#XzZNk`B*fa(PmY(YS5dx
zlQ~qkYc8wyp<0FEiPsF2dlQCSvp9WFnKvs$mfJT7#=zPbH<8ybp+-kNQnzNM5a
zTll!c2vD{RSIO%}4g}p+)^6?K)H3y!l`Dgo@3m;T_0SFhIpr?Zc115~YC%_Eizi9}
z>bp;V`AOctP4{p6uQVRI4UuZ3iZV3y9(5YD;2Y072DPE_e8+^|F`1AqJkBxR$@>a9k#p~7|Z
zZIe}7{8*(S0c4bJ0ANUY?+V|$%=a$6_hek-d-MFH&QEIPoBHto6K}=DM?P8gPOR4Y
zPCgtu6XN-fHLu0m#5!;G$n$NCzS%{czo?aOb9draQ!`R_R!y>P0uW#Zat}x!#xy`7
zL$!I;cTyL)TB87c7|qhAh8ZR5Rw8?>xoXukP+^8m>8Pgpy5mr*%z>5q>o~T*&sItv
zgx8AlrW@!$4V0*{*^#vIS|SH
zgn4m_=4e{-aB#B2Y?
z>lg{RzX@^x|DRD|TIgkE<{!$a@);nyKA%#>bQVyoyouHGtrp&7sz^ob6#%cnp8d!o
z{r@fOxe|;l2P64lpC0T(1~o}j`V=68_9KJVHdIh(uEe10W69^>-Rg&RPW+IKMH9sI
z>2HlmgfkK_qD-s84d5?lEeNj#+sVIw(H}
zzo04x0X})a6J6tzcNW^V6rIpQn<>B`0QjSpIw)81;P`4s
zF9u4O*IzHhQP%^9u*Yk?v2_+o5s#{#=DbAKdf+h8i>iXFzSw%yZvSc%fTzBZ9gsY;
zJC3^D&l)&@hHWZS-&+0?bggykYU9q-MSBfZ0|!7S
z1c#P`LrdrK!CiWASFy3&mlB}h1(JG&pi);aNBLwUETtJn`RSn}E%v7^jw7w^r<*x|
zhP$*$ug?)*Lm#V~zlI-%s$*8`D}RxDpRW4ryZGxEyA9*J4{hTzGABeNFobJHY3*q#
zBV*8mI@z#u?8E#Y15gmR6qk_kd~!Z)S?)RqgMWqH5U};OdtyZbplZFp99?pJ?aN2@
z>5+Ypo0lU8waCG>j&;#p?n?W=Kq9XY(Bj5`F84;Lbh#Jz9`V?pdK^c5?xz6`pv5i0
zZ5Q2#UxtkJfEv~jRJnOsO=@`^ueDVlXahde3H9+OiTqQb2O=jx0tc8DNmR`BsWCY6
zj{tRkyLCzGMclsxJOw+$S);$@&R#g{$@>R%|G?7jW&e=oAN#HS(b2yR0?;3e>
zc6ryB#*dj>c=npm+Xa6*A*+zFvc0mUdP;8Mu?Bl#KxS8szXbQk8?LaGi~
zROYEbEmNjfndehIqN$odiFFac55mJ8Hp)^E<$Tm5_dun2tnoJH&|Q$U5a7I0n8}K>Y0P^C;e**k+H8?=
z_6^77Y&I?94sIQ{hC+u-4V623N2=jidTjPV2-_>>P}4qax$H6r_$MkY6h|nW96O{%!$%e0VqK^SzL4afFe!!#$
zt8bgJ0sL-4Mu3c^JOWk)`&9lFZ2{+7_s@eL;e2aJrOA^M-~QpV37XaZ1j3cjvOb6q
zS5wi(IlExloLw*MP`q`)tZ+G<134eqCnp39te~mc2`4-NmIfZZ`87mCA^YtC?R-Li
z_cASL8JVWFU2Y=!34ql=`~BcuU(tyba4gk%|Fj<5QFLRGC!4!-
zU3&YliQYgOnu?9(j{JcsXap34Qz^wE{aAM5*0n+0GlGgAk;IA@&pjdQt=
z9{KeC!-y2;v@;j9i&OfU=~7X-Zf5F0SECacAHe-9x!KRMH2WCFQLe6=>u7Ys&J#ec
zeW~{g2ysL9@rfsw^h1+0zdq;)IYajyMFL4J7;-sx=T0n5EnRq|EM3wgKQ0oec>>6g
ztPcVXXKd+5a18DQi)95~&Uu{60dHYi2LhjfN4pU)WfoaD8kED37%lLDO>$P2;S-dY
zd4i&tV&cG7mpXS1zG0zaXNnLN+IEFTdSj_a1m@;xa|wM`O{G)nT#SCSBjCZW`Jqu?P5X?bCg^F7`6)yB^xsq!
z_-+r0`JMxi|DHc&YNEsPEELf&DgOolP6jxxKmuC%E|3FSy}LkmXzSmrWRGUO3*@A>
j{#_u$n)NP_w>9fsbX~S_u#olPoqaE8%d2(T>8t-IAI>2+
literal 0
HcmV?d00001
diff --git a/scaler/client/__pycache__/object_buffer.cpython-311.pyc b/scaler/client/__pycache__/object_buffer.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..416e0854e6f369863daab6276ae151e370fa2a32
GIT binary patch
literal 7221
zcmcgxTWk|o8lIb-u``KH;v^Uo!X;1{sG)&eSOwG;mU3&-7FvPZFgTt`aOz8XW+Ix%
ztdN#n<)NzRsuGn}mAq1yrAwtvB_8N%rIq%HmzBoRRIx^igtXd+z8S&3@Y4PNGoG=>
zNdWD`9-sVk&Y6GC>|z;}ndx?>#xZX(O5i4=1n+PRa}WOPbO#>y
zoC=fdp?s~tCwcjjn-%sv6L0$odB@^Di+>|OZ;8iK*=%Yut6IXjjHX-S_^h5O)(W0iNhX!eET{x(ZRh9n>EnfbUQO!-Z0Js3oy&hL>jgOrpUZ_Qd2(J?HDn#1
zoB=Li*3~@rw8?Hbo!4|Sm&Q@zoYgU{=2eo?Rk=Xf<&0wWx^0>}cKvnQT`Ls$TII9_efb7iGj-Q3`{mUTF-GgOu
z$P|YRai|jNGsHeyyxdei-6#rIv1#C3)?dT|jT;+h%gXq2(t)k5Ef61ExkK@Zwt
zAb{JR*h)`gxhHW~TRKthIb!x4S(H2y{-`VtnBssT4pc%hLyXzT4ca|{3Dnj{upK6V
z)LR)M&srqARkw0Tt6ftkmOt`9R0%WK#X>M)1_
zm9>9G+_x<5yL+ZA4x8ez;bgjEj|oIiL2(3;C&f{Hl^TchSJ<0i80)Pkn*7eon8QY|
zHG`W@lkR#VhM94KXlBN3g;8B^r*Cq;0^WKLo%Plt`#JMDa|hOP&()J=6zg4eey>q)
z^uM_3I5*}I6HX5YdxCE0HZu((ndpY4vvt8h*@Xp^U1K=Kf5Cm$13VHz!C72oa>|^%Mi=`SrScTXVyG%Ifws>=E8^B=aqFto
zYf8IU!#m9Iuo*s92}izcUk&$~;oX1K?v5L;j2hvik6VPc_NNSx$GkrP-2kZ~`A?YH
zvHK7HZ~QbQ0b}mia}TIo-Rs^!_g-wN|1X-=p#5HrYN|a~_Zn?TDF_?1CV4fo(R#L0
z>q!uSb6fmz&i{
zKQLMTT3h(k>99^8G=+HUv0TEwehUQbPPEsI##f?;
zmZOK3-YiFl&FJuA`)XuQN&n_oOTQ{dPMVRE0H$bvMe16WI&Z!CRnX`^e)rN}CjLC}
zt^A$*;JpW_2k)7$z7DN+E=wm&>BNe3c3C=Gmfkj{w^yZ|rDJ7jz?25SK5YT#24u0l
z8f3!TTyN&$1?J^ipZ)(U8u<;;$TwLep~#U)kt?}23&wZT9Vp#%Yx&m{wkVVuaPGcp
ziewdtcbl-uf6c$l7}tF|qNbLLLeb~3giWU@FFwKY31bVsQ3!>$XP*81^UvfX=qD~x
zT#^YWlR6@=U=?qQ>>E3ZLVxoatKE*OmVekbB5KLzfKk
z619x>GcSl4-g7%B-LqeSa9+?Q@hhRug5Xl4u0~cC4A$ZA45Ygmy
zB2(N>txba>+;D$Y&hI-?>Vo*tcVy`deLOr$AI&5z
z__p6|tupXvCX(RWR{9iMw=JEek7fcKTd{3xDNP?sn;-1D-%E=RAy#qTHk2gRHv(to
z!S5Czl<|!;6MxwEN|kX*zvO$h%D7}J?~_@aEI!wvLD8NtJaEGBXeJTCmxM2o+EY%l
z^*JY5d(sYv&9|1?dw}APN3B*kW2e&Dl%}Z~`2gB1zdk<;mzMS}(@jTvK=<3;Wpvk8
zpGGgs6ZS26qV5tBxy+}Tyhd?kZLV=;3JDGVu~EMwr3kY&)nCu)nJiogQ)glI*4mu_
z2d>@qtA%_51JOd>J|Jz|fG4MFT54Lg&s|POwcZmoZFdHv#sUK$E8vR3Mmz;4y52fq
zpNXUPKOl~gPv8XxiQ4Nxt31oH6{f>*a)oKTL4Ory(0IB0$oUpJdBhAE-dte@3~#P5
zhYfG8FvEs7SC~=5n=8zR#(KHxXSg<#?XGe>3y@n&9%I{6B#(K9>v377cUkzSU6tAR
O@z{R%^f@e4!v6)5*lvda
literal 0
HcmV?d00001
diff --git a/scaler/client/__pycache__/object_reference.cpython-311.pyc b/scaler/client/__pycache__/object_reference.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..337c07c764035731251961e9dc0213f7a6f9edd1
GIT binary patch
literal 1653
zcmZuw&1)M+6rb5IeOU4q%dry*jZ#}})CMcQ6%zMQN{et&i$!}-!NXcJme*bFI?5ik!P_VB#OnyQlkovH
z6pHJpto?
zM}_D769DKN@1bpw;`Lmi3Y-&vIID0@I)4te{hor{daZk_-n1OXljQeB?v<|oplH*T
zZt6i%q+Yqe<+92J+1wBy(jQ}}i#eOV$Wy%ds^&{YbLK!1|h>d<_xtVG(D
z7
zI^1n0_XjpHld|jv2oq)F;ELz5gnT4N!JiPVd?my$9K=0heh1w7ZGg$zUH10cg?>)>&jY{^7UuhpetmS%IW3e1g;Hpi!gNViS}W(o
zJcuugkSI$?+oK&v$T>oubS!6N(KmJj=0_$Wmg{7|CRka(PJ-#BKRM0ral?US-YRYH#(~>X|tkYT|#Y&7NiRy=t%etUA=N
zPB5^4i3Zjeg<=GTrY|3SF+{K(26CH(9M(S&FHAyQqA04YMpM)Ztfpi6K6IT)4sr2|
z;B(o~Mf!0}+NbHnR!cTxZV
literal 0
HcmV?d00001
diff --git a/scaler/client/agent/__init__.py b/scaler/client/agent/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/scaler/client/agent/__pycache__/__init__.cpython-311.pyc b/scaler/client/agent/__pycache__/__init__.cpython-311.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..1a622c999c14f44adddb06738b7e226ce9f9b288
GIT binary patch
literal 183
zcmZ3^%ge<81fG0R=^*+sh=2h`DC095kTIPhg&~+hlhJP_LlF~@{~09t%S%5aKQ~oB
zJtwt1vsgbTGcUb7F*Ck0Gd?XfFFrj#FTJWXU%xmxF(
+
+
+
+ 
+