@@ -42,6 +42,11 @@ Documentation is built with Sphinx from `docs/` (`pip3 install -e '.[docs]'`, th
 `make -C docs html SPHINXOPTS="-W --keep-going"`, see [docs/README.md](docs/README.md)). Pull
 requests that touch the docs must build without warnings: CI runs that same command.
 
+Use the Python interpreter that has this checkout installed in editable mode (`pip show
+lithops` shows the *Editable project location*). Before reporting a change as done, run
+`ruff check .` and the localhost test suite. Run long jobs (the full test suite, Docker or
+runtime builds, docs builds) in the background.
+
 ## Repository map
 
 | Path | Content |
@@ -75,6 +80,9 @@ requests that touch the docs must build without warnings: CI runs that same comm
 2. **Every backend looks the same to the core.** Backend-specific code stays in its backend
    package; the executor, invoker, job and worker code must not special-case a backend by
    name when an interface method or a config value can express it.
+   Backends are self-contained: a new backend starts as a copy of a similar one, and logic
+   may be duplicated between backends. Do not refactor backends into shared helpers or base
+   classes to remove that duplication.
 3. **The worker must tolerate the environment it runs in.** Worker code runs in minimal
    runtimes (Lambda, containers, VMs, macOS/Windows localhost). Imports of optional
    dependencies stay lazy, and a failure in the user's function is reported in its status
@@ -114,8 +122,8 @@ requests that touch the docs must build without warnings: CI runs that same comm
 ## Safety
 
 - Do not run tests, examples or `lithops` CLI commands against real cloud backends (AWS, GCP,
-  Azure, IBM Cloud, Aliyun, Oracle, Kubernetes clusters) and do not build or push runtime
-  images without the maintainer's explicit approval: it creates billable resources and uses
-  real credentials.
+  Azure, IBM Cloud, Aliyun, Oracle, Kubernetes clusters), do not build or push runtime images,
+  and do not publish content without the maintainer's explicit approval: it creates billable
+  resources and uses real credentials.
 - Never commit credentials or configuration files (`~/.lithops/config`, `.lithops_config`).
 - Do not commit or push unless asked to.
@@ -4,15 +4,24 @@
 
 ### Changed
 - [Core] A storage or monitoring service at a loopback address is refused for serverless workers, which cannot reach it; standalone and Singularity workers get a warning.
-- [Core] `RetryingFunctionExecutor` lists a bucket, prefix or directory before invoking, so that each object is retried on its own, and rejects `obj_chunk_size` and `obj_chunk_number`.
+- [Core] `wait()` timeouts no longer arm a SIGALRM: they are checked between polls, work from any thread and leave the caller's signal handler and alarm alone.
+- [Core] An `obj_chunk_size` no longer than `obj_newline` is raised to one byte more.
+- [Storage] Glob patterns in the input objects work on every storage backend: a key with `*` is a pattern, as it already was on AWS S3 and IBM COS, also from a Windows client.
+- [CLI] Times are shown in the local time zone without `pytz` and `tzlocal`, which are no longer used.
+- [Standalone] `lithops worker list` shows the last error of each worker and how many of its processes are busy.
+- [Knative] Updated to current Knative Serving (tested with 1.23): one container per call in flight with the current autoscaling annotations, `kubecfg_context`, `namespace` and `runtime_arch` keys, the Contour networking layer, cluster-local calls from inside the cluster, a deploy that updates an existing service and reports why its revision fails, and a default runtime that runs as a non-root user.
 
 ### Fixed
-- [Core] Chaining, cleanup, invocation errors, generator inputs, empty `fs`, `wait()` timeouts (threads, SIGALRM, `ANY_COMPLETED`), argument binding and `RetryingFunctionExecutor` retries and percentage waits.
+- [Core] Chaining, cleanup, invocation errors, generator inputs, empty `fs`, `wait()` with `ANY_COMPLETED`, argument binding, and `RetryingFunctionExecutor` retries and percentage waits; a failed object, or chunk of one, is retried on its own.
 - [Worker] Partitioned object streams (lost, duplicated or cut rows, rows longer than 128 KB, multi-byte `obj_newline`, `readline()`, iteration), module discovery of nested functions, a parameter named `future`, and `extra_env` restoring the previous environment.
-- [Localhost] Absolute interpreter paths are accepted as runtime; storage listing is a literal prefix match including dotfiles, and open-ended byte ranges are supported.
-- [Storage] `CloudFileProxy` path handling (`isfile`, `listdir`, `exists`, leading `/`), double `close()`, and `r+` mode writing back.
-- [Monitoring] Worker tokens of jobs dropped by `wait()`, replaced monitors and restarted invokers are handed back once, a held `__end__` is no longer replaced by an `__init__`, and the workers of a job that owes no token are forgotten instead of kept for the life of the executor.
+- [Localhost] An absolute interpreter path with any character, such as `@` or a space, is accepted as runtime; storage listing is a literal prefix match including dotfiles, and open-ended byte ranges are supported.
+- [Storage] `CloudFileProxy` path handling (`isfile`, `listdir`, `exists`, leading `/`), double `close()`, and `r+` mode writing back; Swift `generate_bucket_name`, Aliyun and Oracle listings past 1,000 objects, and open-ended byte ranges on GCS, Azure and Aliyun without modifying the caller's `extra_get_args`; glob patterns outside the bucket root on AWS S3 and IBM COS; Swift listings return `Key` and `Size` like the other backends and go past the first page.
+- [Serverless] Oracle Functions reports a missing `oracle_f` section, an Aliyun zip runtime builds without Docker, and Kubernetes no longer waits for ever on a runtime image it cannot pull or a master pod that does not start.
+- [Standalone] A failed `init()` on AWS EC2, GCP Compute Engine and IBM VPC now deletes every resource it created, including those whose tagging, attaching or rules failed; IBM VPC no longer uses the deprecated `code` of SDK errors; the `vm` backend loads its documented configuration, and Azure `request_spot_instances` creates Spot worker VMs (now `False` by default: B-series sizes, like the default worker, do not support Spot).
+- [Monitoring] Worker tokens of jobs dropped by `wait()`, replaced monitors and restarted invokers are handed back once, a held `__end__` is no longer replaced by an `__init__`, and the workers of a job that owes no token are forgotten instead of kept for as long as the job monitor runs; the shared RabbitMQ connection of a worker is closed at exit.
 - [Concurrent] `wait()` and `as_completed()` accept generators.
+- [Core] An executor that fails to be created no longer logs a misleading `job_monitor` error at exit.
+- [Knative] Calls longer than 5 minutes are no longer cut off by Knative, a load balancer ingress is reached at its own port and may be a hostname, `https` service URLs work, a failed call is reported instead of being invoked again for ever, and a service that never gets ready raises an error.
 
 ## [v3.8.0]
 
 
@@ -76,6 +76,9 @@ not exercised in CI, so mention in your pull request which backend you tested on
 - **Backends:** backend-specific code stays in its package (`lithops/serverless/backends/`,
   `lithops/standalone/backends/`, `lithops/storage/backends/`). Import the provider SDKs
   there, not in the core modules, so Lithops keeps working without the extras installed.
+  Each backend is self-contained: to implement a new one, start from the code of a similar
+  backend and adapt it. Duplicating logic between backends is fine; do not factor it out into
+  shared modules or base classes, so that a change in one backend cannot break another.
 - **Configuration keys:** document new or changed keys in `config/config_template.yaml` and
   in the backend's page under `docs/source/compute_config/` or `docs/source/storage_config/`
   (general `lithops` keys in `docs/source/lithops_config_keys.csv`).
 
@@ -118,6 +118,7 @@
     #master_instance_type: t3.micro
     #worker_instance_type: t3.medium
     #delete_on_dismantle: True
+    #worker_create_retries: 2             # Attempts to bring up each worker VM (create/reuse modes). Minimum 1
     #max_workers: 100
     #worker_processes: AUTO               # Default: number of CPUs of the worker VM
     #ebs_volumes:                         # Optional list of EBS volume mappings
@@ -165,7 +166,6 @@
     #min_workers: 0
     #max_workers: 100
     #worker_processes: 1
-    #invoke_pool_threads: 100
     #docker_server: pkg.dev               # Marker for Artifact Registry image names
     #artifact_registry_repository: lithops
     #runtime_include_function: False
@@ -192,6 +192,8 @@
     #ssh_key_filename: ~/.ssh/id_rsa
     #request_spot_instances: False
     #delete_on_dismantle: True
+    #worker_create_retries: 2             # Attempts to bring up each worker VM (create/reuse modes). Minimum 1
+    #worker_public_ip: False              # Also give worker VMs an external IP (default: Cloud NAT only)
     #max_workers: 100
     #worker_processes: AUTO               # Default: number of CPUs of the worker VM
 
@@ -245,8 +247,9 @@
     #ssh_username: ubuntu
     #ssh_password: <RANDOM_IF_NOT_PROVIDED>
     #ssh_key_filename: ~/.ssh/id_rsa
-    #request_spot_instances: True
+    #request_spot_instances: False       # Spot worker VMs; B-series sizes do not support Spot
     #delete_on_dismantle: False           # `True` is not yet supported on Azure
+    #worker_create_retries: 2             # Attempts to bring up each worker VM (create/reuse modes). Minimum 1
     #max_workers: 100
     #worker_processes: AUTO
 
@@ -319,6 +322,7 @@
     #master_profile_name: cx2-2x4
     #verify_resources: True
     #delete_on_dismantle: True
+    #worker_create_retries: 2             # Attempts to bring up each worker VM (create/reuse modes). Minimum 1
     #max_workers: 100
     #worker_processes: AUTO
     #singlesocket: False
@@ -435,21 +439,21 @@
 
 #knative:
     #kubecfg_path: <PATH_TO_KUBECONFIG>
-    #networking_layer: kourier            # One of: kourier, istio
-    #ingress_endpoint: <INGRESS_URL>
+    #kubecfg_context: <CONTEXT_NAME>
+    #namespace: <NAMESPACE>               # Defaults to the namespace of the context, or 'default'
+    #networking_layer: kourier            # One of: kourier, istio, contour
+    #ingress_endpoint: <INGRESS_URL>      # Found automatically when not set
     #docker_server: docker.io
     #docker_user: <REGISTRY_USERNAME>
     #docker_password: <REGISTRY_TOKEN>
     #docker_namespace: <REGISTRY_NAMESPACE>
-    #git_url: <GIT_REPO_URL>
-    #git_rev: <GIT_REVISION>
     #runtime: <RUNTIME_NAME>
     #runtime_cpu: 1                       # vCPU
     #runtime_memory: 512                  # MB
-    #runtime_timeout: 600                 # Seconds
+    #runtime_timeout: 600                 # Seconds; at most Knative's max-revision-timeout-seconds
+    #runtime_arch: <amd64|arm64>          # Defaults to cluster nodes; falls back to amd64
     #max_workers: 100
     #worker_processes: 1
-    #invoke_pool_threads: 100
 
 #openwhisk:
     #endpoint: <OW_ENDPOINT>              # Mandatory
@@ -590,7 +594,7 @@
 #prometheus:
     #gateway: http://localhost:9091       # Pushgateway base URL. Default: http://localhost:9091
     #job: lithops                         # Pushgateway job label. Default: lithops
-    #instance: <HOSTNAME>                 # Pushgateway instance label. Default: the hostname
+    #instance: <HOSTNAME>                 # Pushgateway instance label. Default: $LITHOPS_TELEMETRY_INSTANCE, else the hostname
     #username: <USERNAME>                 # Basic auth user, when the gateway needs one
     #password: <PASSWORD>                 # Basic auth password, when the gateway needs one
     #timeout: 10                          # Push timeout in seconds. Default: 10
@@ -603,6 +607,7 @@
     #endpoint: http://localhost:4318      # Collector endpoint. Default: http://localhost:4318
     #protocol: http/protobuf              # One of: http/protobuf, grpc. Default: http/protobuf
     #service_name: lithops                # OTel service.name. Default: lithops
-    #instance: <HOSTNAME>                 # OTel host.name. Default: the hostname
+    #instance: <HOSTNAME>                 # OTel service.instance.id and host.name. Default: $LITHOPS_TELEMETRY_INSTANCE, else the hostname
     #headers: {}                          # Extra headers, e.g. authentication
     #timeout: 10                          # Export timeout in seconds. Default: 10
+    #export_interval: 3600                # Seconds between the OTel SDK's own exports, a fallback to telemetry_interval. Default: 3600
@@ -18,7 +18,7 @@ Primary Executors
 * **RetryingFunctionExecutor** (``lithops.RetryingFunctionExecutor()``):
   A robust wrapper around ``FunctionExecutor`` that transparently handles retries on failed tasks.
   It supports the features of ``FunctionExecutor`` with added automatic retry logic, improving fault tolerance and reliability for unstable or transient-failure-prone environments.
-  A failed call is retried with its own input: a bucket, prefix or directory passed to a function that processes objects is retried one object at a time, and splitting objects into chunks (``obj_chunk_size``, ``obj_chunk_number``) is not supported.
+  A failed call is retried with its own input: a bucket, prefix or directory passed to a function that processes objects is retried one object, or one chunk of an object, at a time.
 
 Secondary Executors
 -------------------
 
@@ -111,7 +111,9 @@ Open an ssh connection to the master VM (Only available for standalone backends)
 ``lithops worker list``
 ~~~~~~~~~~~~~~~~~~~~~~~
 
-Lists the available workers in the master VM (Only available for standalone backends)
+Lists the available workers in the master VM (Only available for standalone backends). For each
+worker it shows its status, how many of its processes are running a task (*Busy*, for example
+``1/4``), the last error it reported, and the time left before it stops itself (*TTD*).
 
 +------------------------+----------------------------------------------------------------+
 | Parameter              | Description                                                    |
 
@@ -123,6 +123,7 @@ In summary, you can use one of the following settings:
 |aws_ec2 | master_instance_type | t2.micro | no | Profile name for the master VM |
 |aws_ec2 | worker_instance_type | t2.medium | no | Profile name for the worker VMs |
 |aws_ec2 | delete_on_dismantle | True | no | Delete the worker VMs when they are stopped. Master VM is never deleted when stopped |
+|aws_ec2 | worker_create_retries | 2 | no | Attempts to bring up each worker VM in create and reuse modes. A worker VM that is not reachable over SSH in time is deleted and created again until this many attempts have been made; then that worker fails and the job runs on the workers that came up. Minimum 1 |
 |aws_ec2 | max_workers | 100 | no | Max number of workers per `FunctionExecutor()`|
 |aws_ec2 | worker_processes | AUTO | no | Number of parallel Lithops processes in a worker. This is used to parallelize function activations within the worker. By default it detects the amount of CPUs in the `worker_instance_type` VM|
 |aws_ec2 | runtime | python3 | no | Runtime name to run the functions. Can be a container image name. If not set Lithops will use the default python3 interpreter of the VM |
 
@@ -51,6 +51,8 @@ az provider register --namespace Microsoft.OperationalInsights --wait
 az containerapp env create --name lithops --resource-group LithopsResourceGroup --location westeurope
 ```
 
+Lithops deploys its container apps in the `region` of the configuration, so create the environment in that same region (or set `region` to the region of the environment). If Lithops cannot find the environment, the runtime deployment fails at once with an error naming the environment, resource group and region it looked for.
+
 ## Configuration
 
 1. Access the [Storage Account](https://portal.azure.com/#view/HubsExtension/BrowseResource/resourceType/Microsoft.Storage%2FStorageAccounts)
 
@@ -89,7 +89,9 @@ Edit your Lithops config and add the relevant keys:
 |azure_vms | ssh_key_filename | ~/.ssh/id_rsa | no | Path to the ssh key file provided to access the VPC. It will use the default path if not provided |
 |azure_vms | master_instance_type | Standard_B1s | no | Profile name for the master VM |
 |azure_vms | worker_instance_type | Standard_B2s | no | Profile name for the worker VMs |
+|azure_vms | request_spot_instances | False | no | Create the worker VMs as Spot VMs. They are evicted when Azure needs the capacity back. B-series sizes, like the default `worker_instance_type`, do not support Spot |
 |azure_vms | delete_on_dismantle | False | no | Delete the worker VMs when they are stopped. Master VM is never deleted when stopped. `True` is NOT YET SUPPORTED |
+|azure_vms | worker_create_retries | 2 | no | Attempts to bring up each worker VM in create and reuse modes. A worker VM that is not reachable over SSH in time is deleted and created again until this many attempts have been made; then that worker fails and the job runs on the workers that came up. Minimum 1 |
 |azure_vms | max_workers | 100 | no | Max number of workers per `FunctionExecutor()`|
 |azure_vms | worker_processes | AUTO | no | Number of parallel Lithops processes in a worker. This is used to parallelize function activations within the worker. By default it detects the amount of CPUs in the `worker_instance_type` VM|
 |azure_vms | runtime | python3 | no | Runtime name to run the functions. Can be a container image name. If not set Lithops will use the default python3 interpreter of the VM |