AppScale 3.8.1

Highlights of features/bugfixes in this release:

• Fixes a regression in the 3.8 release that prevented push tasks from being executed for secure URLs.

AppScale 3.8.0

Highlights of features/bugfixes in this release:

• Login continue scheme
• Dispatch Routing from yaml
• Install newer psutils to handle kernel 5.X output of /proc/diskstats
• Allow custom service accounts for restore operations
• Remove Cassandra-specific backup/restore scripts
• Fix usages of ZKTransaction class
• Update bootstrap.sh usage text for agents
• Stop unassigned services
• Remove Fixnum warning
• Remove old zk nodes structure
• Remove outdated scripts
• Display a selection menu for available accounts
• Remove support for Debian Jessie
• Search backup restore
• Store FoundationDB clusterfile content in Zookeeper
• Use monotonic module when measuring intervals
• Limit connection retries when stopping celery
• Support custom service accounts for datastore_admin backups
• Install FoundationDB client libraries
• UA Server backup/restore
• Add a rough interface for managing service accounts
• Remove M2Crypto dependency
• Retry applying transaction mutations
• AppController haproxy reload via monit
• Escape property values when outputting table
• Be more flexible when selecting a composite index
• Remove ntp dependency
• Flatten path before extracting the entity group
• Check if we have flower, and added dependency
• Remove iaas system manager
• Disable ssl verification for Java application server channel client
• AppController command output to standard error / logging verbosity
• No need to use sudo during installation
• Search2 integrated
• Datastore service assignments calculated by admin server not controller
• Delete SDK from AppTaskQueue
• AppController update jsp jar when rebuilding java application server
• Java 8 runtime
• Add initial FDB datastore backend implementation
• Handle scheme redirects with the dispatcher
• Fix non-cloud mode of AppScale launch.
• Perform datastore operations in batches
• Scaled nodes changes
• Stop services concurrently
• Higher TaskQueue e2e test stability
• AppServer generate internal admin login secret where needed
• Add initial implementation of datastore-backed sessions
• App cert
• Move of appscale.tools.agents to appscale.agents - appscale repo.
• Fix python detection in taskqueue tests
• Better venvs support in AppController
• Resolving dependencies between tq and adminserver
• AppServer admin server module should not be enabled for python runtime
• Do not modify /root file mode on install
• Remove support for trusty
• Increase the monit safety sleep time
• Limit version of cassandra-driver
• Dont override hostname
• Use pymemcache to implement the memcache API
• Remove local_ip helper function
• Splitting service specific and service common code
• Retry 3 times with backoff on socket err in AppServer
• AppServer request environment setup
• Remove old soap calls
• Fix the Ruby client's get and set property methods
• Blobstore application id header and relative success url
• AppServer ensure request type set for request
• Add check for system stats
• Use the Python API server for some Java runtime calls
• Removed cached local IP
• Pass app identity calls to the external API server
• Datastore service index manager zookeeper callback state parameter
• Avoid using localhost when waiting for service
• Use output of appscale-admin summary
• Check for more results during query
• Remove haproxy statistics page
• Clear monit environment variables before exec
• Enforce min machines
• Cron schedule list parsing and sub-daily interval crontab generation
• Update default config for java runtime
• Move protobuffer handling to different level
• Remove HAProxy stats URL
• Wait for result before setting chosen task name
• Add faceted search support to the Python runtime
• Continue batch query if limit hasn't been reached
• Restart unavailable instances
• Allow push queue bucket size and max concurrent requests parameters
• Datastore entity locking resolve deadlock retry syntax
• Async coroutine retry for datastore large batches
• Ensure application instances logger tag is 32 characters or less
• Datastore statistics timestamp second resolution
• Configure open file limit for monit
• Add B* instance classes
• Proxy stats 7x performance improvement
• Add aws_subnet_id controller option
• Allow "key" along with other equality filters
• Remove "for_cas" field requirement on set
• Access AC and AdminServer with private IPs
• Remove dead code from helper library
• Only list unique items in project list
• Prevent SDK from starting more than one instance
• Remove login role

AppScale 3.7.1

Highlights of features/bugfixes in this release:

• Update cron when login property is changed
• Remove --force-private from fast-start.sh
• use dns for ec2 since it resolves to the private ip for marketplace
• Other clouds will use public ips for login property

AppScale 3.7.0

Highlights of features/bugfixes in this release:

• InfrastructureManager restructure
• Change jobs to roles
• Raise API server errors as Java exceptions
• Don't add tag data to task if the task has none
• Rescue Errno::EHOSTUNREACH
• Add a timeout to the health check request
• Retry dashboard deploy operations on timeout
• Remove monit maintenance port from the firewall.
• Add required ProjectGroomer argument
• Remove index management from datastore clients
• Add standalone API Server from SDK
• Update allocate methods in datastore stubs
• Replacing zookeeper_locations.json with zookeeper_locations
• Bound update threads
• Use initial flags from increment request
• Add partition list to RabbitMQ stats snapshot
• get_index method reorganizing
• Made appscale.common python3 compatibility changes
• Drop requests that don't match the login property
• Use login property when setting cron locations
• Ensure requests to channel path match
• Use login property as ejabberd host
• Serving status modification for application versions with manual scaling
• Route instances as soon as they are registered
• Smooth current sessions for autoscaling
• Don't wait for autoscaled nodes
• Add support for the target field for cron jobs
• Rework search query parsing and building
• Limit the number of entities fetched per statement
• Store hints within the persistent state directory
• Communicate with ejabberd using private IP address
• Accept multiple values per field in SearchService
• Handle datastore connection errors with TQ server
• AdminServer application deployment check now ignores working directory
• Master reload state
• Postgres connection pool with autoreconnect
• Accept single exception class as retry_on_exception value
• Terminate not started AppServers first
• Check for valid local source archive
• Project ID validation upgrade
• Application scaling for version configuration
• Added service ID validation and modified version ID validation.
• Implement apps.services.list
• Exit early if version is not found in zookeeper
• Use zookeeper for stopped versions
• Ec2 private ips
• Defer to vendored framework for endpoints 2
• Accept all arguments to _handle_request
• Move blob fetching work to DatastoreBlobReader
• Use a fallback mechanism on connection failures
• Test query that exceeds Cassandra page size
• Use gossip to find live Cassandra nodes
• Collect each page of results within callback
• Improve speed of rebalance script
• Move instance fulfillment work to InstanceManager
• Allow change instance type
• Add cassandra status to Hermes
• Fix snapshots caching in Hermes
• Force the push worker to time out stuck tasks
• Optimize Service Stats
• Make communication with rabbitmq async (in Hermes)
• Fix list_tasks method in TaskQueue
• Taskqueue load test
• Taskqueue e2e test
• enable-datastore-viewer.sh: Fix command extracting admin port
• Add IndexManager
• Re-add editing capabilities to datastore viewer
• Add basic pull queue viewer to dashboard
• Logging update to use per-module loggers
• Run datastore servers in a custom slice
• Add support for Bionic
• Add support for Stretch
• Remove support for wheezy
• Fix bionic php dependencies
• Use the unittest module to run end-to-end tests
• Use a patched version of Cassandra
• Decrease Cassandra heap size when sharing roles
• Install Python MySQL client during build
• Update Python and PHP runtime to 1.8.4 SDK
• Pin SOAPpy to version 0.12.22
• Require attrs>=18.1.0
• Ensure newer version of PyYAML is installed

AppScale 3.6.1

Highlights of features/bugfixes in this release:

• Issue-715 - SSH login problems when starting cloud-mode on GCE (tools)

AppScale 3.6.0

Highlights of features/bugfixes in this release:

• Close PG connection when cleaning up project
• Catch QueueNotFound error in TQ REST API
• Use security token when making Monit API calls
• Add E2E test for queries involving ID separator
• Error handling for starting or terminating instances
• Handle UpdateIndexes calls
• Add appscale-admin restart command
• Use hmac helper for authentication check
• Stop services in a particular order
• Handle ID separator in key name
• Do not delete push task names immediately
• Fix nginx config for handlers with secure field
• Reduce diffs with the 1.8.0 SDK (Part 2)
• Autoscale logs
• Remove BR service
• Do not validate SSL certs for SOAP services
• PullQueueus on postgresql
• Copy PATH variable from controller environment
• Detect missing row when applying transaction
• Install older tornado when Python is too old
• Add transactional counter test
• Remove indexless sorted ancestor queries
• Counts only ESTABLISHED connections
• Reject lease modifications that specify wrong ETA
• Controller application upload error output now escaped
• AppManager should not re-extract source for an application revision
• Determine new_eta for leased tasks later
• Use ranges to keep track of merge join queries
• AppDashboard redirect cleanup
• Java app server disable TLSv1 protocol
• Use yaml.safe_load rather than yaml.load
• Simplify rollback_transaction method
• Install headless OpenJDK package
• Remove sshguard
• Java AppServer javac build options update for version and debug
• Continue preparing archive despite being a hoster
• Limit version of soap4r-ng
• Disable TLS for ejabberd's http_bind module
• Add more logging to pull queue operations
• Restart rsyslog so the app logs can be recreated
• Clean up group locks when possible
• Fix blobstore-backed GCS operations
• Use HTTP API to stop Monit services
• Skip Nginx connection change for Docker build
• Define proper Java version for Trusty
• Assign the proper number of datastore to machines
• RM-626 Fetch application archive as soon as it's available in deployment
• Handle NoSuchProcess error as success when terminating process
• Increase open file limit for nginx
• Handle reads with the index past the last block
• Replace stub with TaskQueue client
• Wait for all hosts before creating tables
• RM-550 Hermes taskqueue stats
• Reduce diffs with the 1.8.0 SDK
• Handle timeouts when querying available tasks
• Ensure Celery worker gets restarted
• Allow PHP instances to use external API server
• Skip stats generation on undeployed projects
• AppController appserver check thread should exit when there is no work
• Handle "service" element
• Use external App Identity service for Java apps
• Install OpenJDK 8 on Trusty
• Allow Go instances to use external API server
• Continue starting deployment if system user exists
• Handle exception when fetching datastore servers
• Use the ServiceManager to start datastore servers
• Retry creating system users when AppScale starts
• RM-542 Async datastore (communication with Cassandra)
• Fix the version delete API method
• Give more informative messages for GQL errors
• Fix dashboard app list
• Issue-2731 Increase Hermes concurrency
• Consider the version's previous ports as available
• RM-616: Use project_id not versions_key to list apps
• Retry operations when nodes are not available
• RM-624 Start AppManager on compute nodes only
• Define the Cassandra cluster load balancing policy
• Register instances with ZooKeeper
• Allow Java AppServer to use an external API Server
• Upgrade to Cassandra 3.11.2
• Add option to update all composite indexes
• RM-592 - Ensure haproxy configuration is up-to-date
• Adjustable max appservers
• Simplify datastore error handling
• Pin eventlet version
• Replace usages of execute_concurrent
• Add a scatter property to some entities
• Use datastore server registry to toggle read-only mode
• Handle docker as a provider
• Wait for all load balancers to route instances
• RM-589 handle failure on fetching local stats

AppScale 3.5.3

Highlights of features/bugfixes in this release:

• Issue-2813 - Controller won't start with latest version of soap4r-ng (2.0.4)

AppScale 3.5.2

Highlights of features/bugs in this release:

• Issue-2770 - Groups are not locked during commit

AppScale 3.5.1

Highlights of features/bugs in this release:

• Upgrade to Cassandra 3.11.2
• Handle Docker as a provider

AppScale 3.5.0

Highlights of features/bugs in this release:

• Add datastore viewer to the dashboard
• Improve lb responsiveness
• Transaction groomer
• Add push queue stats to Hermes
• Allow group locks to persist when commits fail
• Allow different instance type per role
• Handle key inequality filter within property value
• Change appengine role to compute
• Whitelist inbound service warmup
• Run apps without application element in app.yaml
• Use prefix for the haproxy app config file
• Prevent ProjectGroomer worker from stopping
• Stick to the duty cycle length
• Capture Net::ReadTimeout when using Net::HTTP
• Haproxy consolidate apps
• Patch for GAE Issue # 12783 - Conditional import of RAND_egd from _ssl
• Don't clear cron jobs during an up
• Remove tcp_tw_recycle option
• Replace reference to deprecated role
• Reduce Cassandra memory if running other services
• Allow the dashboard to be updated
• Differentiate between project IDs and version keys
• Encode entity names when acquiring lock
• Add external API server
• Allow Hermes to check RabbitMQ status
• Fix JSP compilation under Java 8
• Handle missing WEB-INF directory
• Account properly for pending AppServer.
• Handle UpdateCron requests
• Quote environment variables
• Don't wait for taskqueue master to start rabbitmq
• Autoscale with N load balancer nodes
• Set log after restart
• Clear logs when flushing
• Add support for Jessie
• Remove support for precise
• Use tornado coroutines when performing retries.
• Add bin, bout, hrsp_4xx, hrsp_5xx to proxy.frontend include list
• Retrieve inbound services from ZooKeeper
• Start proper epmd service even without RabbitMQ
• Perform ZK operations off of the Kazoo thread
• Keep track of ZooKeeper state with TransactionManager
• Move AppControllerClient to separate package
• Kill instances when they exceed memory limit
• Set up blobstore routing on each load balancer
• Wait after a restore for cluster to be ready
• Access Tokens for Admin API
• Forward add_routing_for_appserver
• Include version headers when making TQ requests
• Do not raise HttpError outside context of HTTP request
• Use custom template for combined app log messages
• Define property name based on rsyslog version
• Return appropriate error when a queue is not found
• Fix pidfile for RabbitMQ in Docker
• Add sandbox function for expanduser.

Artifacts

• VirtualBox Image
• Amazon EC2 Public AMI
• GCE Public Image
• KVM Image