HPE Helion Stackato |
© Copyright 2016 Hewlett Packard Enterprise Development LP
alsek stack by default. It is
possible to add stacks to deploy applications designed for
different platforms (for example, cflinuxfs2 or win2012r2).win2012r2 stack) allows pushing application code to a WinDEA.usbc, a new CLI tool is available on Helion
Stackato instances (alongside the kato CLI tool).All patches released since v3.6.1 have been included with this release.
docker-app-log: Fixes logging for some Docker images.
routes-flatlined: Ensures that any route but the last one is removed after flatlining.
internal-docker-registry-upgrade: Allows Helion Stackato to perform remote registry pulls from Docker Hub v2.
docker-version-upgrade: Upgrades the Docker client shipped in Helion Stackato to v1.8.3.
config-redis-handle-retry: Improves database connection attempts while Redis is loading.
batch-patch-21-12-2015: This batch patch addresses the following issues:
kato op regenerate all command exits while regenerating the encryption key for the Cloud
Controller database.stackato files <docker-app-name> command does not run.http://EXAMPLE.192.0.2.1.xip.io) do not work.stackato user, giving the process elevated access privileges.stackato_rest crashes when config-redis is restarted.In addition, this patch removes the kato node reset factory command.
batch-patch-13-01-2016: This batch patch performs the following actions:
pip-dependencies-installation: Upgrades the version of setuptools in the Helion Stackato base image
to v19.2.
java-cve: Fixes the CVE-2016-0708 critical vulnerability by removing the appropriate software hooks that can compromise your files.
stack-cflinuxfs2: Adds support for installing the CF Linux FS2 stack and other stacks.
windea: Adds support for adding a Windows DEA node to a Helion Stackato cluster.
add-usb: Adds support for connecting services to a Helion Stackato cluster using the Universal Service Broker.
--disk for the push command.disk and the new cf key
'disk_quota'. If both keys are present, disk_quota takes priority over disk.App has not finished staging to App package
invalid, no bits uploaded.open command to reject stopped applications.dbshell command to distinguish between an existing but
unbound service and a missing service within generated error messages.apps command, adding the new Disk column and showing
additional data with unit letters for Mem and Disk.(set-)environment-variable-group commands.(un)set-placement-zone commands.instance memory attribute.stack, and the
time of last upload.push --no-urls option and (and the cf-compatible alias
--no-routes, allowing worker apps to be specified from the command line.-v, --version, -h, --help, and
-? and implemented the options as commands mapping to version and help.inline-relation-depth > 2 to the new maximum of 2.All patches released since v3.6.0 have been included with this release.
docker run --rm to read
the stackon manifest.duplicate key value violates unique
constraint "buildpacks_pkey"stackato CLI client[301584] Force --cafile and --skip-ssl-validation defaults when cmdr
triggers target access before argument completion phase, to have
certs and skip/debug flags ready.
[301559, 301224, 300024] Process option --docker-image in the interactive SyncV2 code path.
More service information: state of last operation ... Added support, and display.
New undocumented command "debug-options" to list the set of --debug options available to developers.
Implemented support for environment-variable-groups (staging and running). Two new commands, "environment-variable-group" and "set-environment-variable-group".
Tweaks to the output of various commands generating listings, for more consistency.
Test suite updates.
Translate websocket issues on open into regular errors instead of
having them as internal error. Extended websocket debugging.
Activate deeper websocket package debug for --debug mgr|ws|data.
Modified internal http package to have the url which caused error on open in the message.
[301493] Memory reporting by the usage command, assume unit [Byte] for v2 API, stackato 3.x, usage key.
[301411] Gracefully handle the two missing attributes (overridden, default_value) CF dropped from the feature-flag entity spec.
[301346] Update internals to the new location for orphan-relations in list results. Key has changed from "relations" to "orphans". Client now supports both, for compatibility.
[301254] Gracefully degrade the listing of service plans when we do not have the permissions to see service-broker information.
[300462] Modified retrieval of route information to prefer direct access from the space over the getting the full list and client-side filtering, for proper fast knowledge of unused routes. Also fixed the same issue in the "delete-route" command.
[300024] Implemented a new option "--docker-image" for command "push" for the specification of docker images as applications.
Made "scp" more robust against the possibility of trying to access an instance which is down or flapping.
[104102] Restart tunnel app after binding a service, to make it fully known. Without restart the app will not see the new service and thus not allow connection.
[106212] Modified the handling of --path option to push command. Now
has precedence of a manifest path: setting. Fixes the issue of a
manifest.yml suppressing the use of --path, even through a default
path:.
[105017] Added support for killing (== restarting) a specific application instance. The new command is actually an option: "restart --instance N".
[105944] Extended "target" command to allow for a --timeout declaration, and making it sticky. The --timeout of "push" now checks for such a sticky value.
[105550] Activated TLS 1.1 and 1.2 protocols for the client.
Tweaked error reporting involving an application rollback. Error is now reported before the rollback, the rollback is done, and the error is reported again, with a "Reminder" tag and that it forced the rollback.
[106151, 105886] Reworked handling of pre-push hooks to be more portable (SHELL, COMSPEC, /bin/sh fallback for unix(-like) environments).
[300114] Tweaked output of "apps" command to not show the columns "Restart" and "Drains" by default, to make it easier to show on narrow terminals. The old and very wide look can be restored via the new option "--full".
Extended the logstream display to recognize a few fixed strings in the texts and highlight them via colorization (warning, error, note).
[106086] Start log-stream for zero-downtime switchover as well, and use timeouts to detect its semi-start and semi-end.
[105916] Fixed spelling mistakes and inconsistencies of "buildpack".
[105899] Extended "target" command to allow for --no-ws|--ws
declaration, and making them sticky. The --no-ws of push now checks
for such a sticky flag.
[OTRS 030026186]: Fix un|bind-service command when talking to the
"tunnel" application on a Stackato v2 target. Application names are
strings, must not be mistaken for object. For S3 targets we additional
check the class of the object.
[104626] Only for Windows, added a 1s delay between closing the temp file used by command "create-buildpack" and its deletion. Done to avoid the OS race where we try to delete the file while it is not yet unlocked. Small tweaks to the progress reporting.
[105748] Suppress SSL warnings for quiet commands like ssh, scp, etc., and when --json is given, to prevent interference with scripted use of this data.
[104652] Drop ".exe" extension from the application name shown in the help.
[104535] Extended the command "create-service-broker" with option "--public". The plans of the newly registered broker are immediately made public.
New command "show-service-broker" to show the information of a single named broker.
Extended command "service-plans" (and "services") to show the broker a plan is associated with, if any.
[104765, 104857] Extended the 'env' command to show the system variables as well.
[104693] Restrict (interactive/automatic) choice of org on login or other places to the set of orgs the currently logged in user belongs to, instead of allowing all.
[104294] Added alias name --placementzone for option
--placement-zone of command push. Extended the manifest handling
to recognize the key placementzone as a variant of the canonical
key placement-zone.
Note
If both forms are specified, the canonical placement-zone takes
precedence.
[300496] Limit inline-relation-depth to 2 for non-Stackato targets.
kato data export for exporting just the Cloud Controller main database (users, orgs, spaces).kato status now faster. Contacts all nodes concurrently.kato patch checks that it is up-to-date before applying patches.kato start controller needs to check for multiple controllerskato patch install attempts to repeatedly update other nodeskato patch updatekato patch install on non-existent patch should return proper error messagekato data import/export unauthorized service offering errorkato data repair_routeskato relocate containers does not maintain all DOCKER_OPTSkato relocate failure: error trying to unmount an AUFS docker nodekato relocate does not work if data is on different devicesentinel download --for-upgrade should download intermediate versionskato op remap_hosts after data importstackato-ssh breaks when SSLv3 disabled/usr/share/doc directory restored/dev/fd if it already exists.app_bits_controller.rb lost in 2014-05 mergeadmin patch command.--skip-ssl-validation bool. Env. Variable STACKATO_SKIP_SSL_VALIDATION. Sticky setting when used with target command.spaces is invoked with option --all, as this mode does not require an org.delete --all is invoked without a current space, instead of crashing.org-users and space-users with support for option --json.memspec validation type (units used, syntax).--all to the command spaces. When used the command will show all spaces across all orgs, instead of just the spaces in the current/chosen org.restage command for applications.--token-file, and report them properly instead of as internal error.--force-war-unpacking, for clarity.--colormode.push (as update). Added generation of sensible output for when 0-downtime switch-over was available and got triggered.--keep-form to commands create-buildpack and update-buildpack. Taking a path argument its use allows introspection of the form-data generated and uploaded by the client.kato node upgrade (Sentinel).gnatsd to 0.5.2.zeromq 3.2.4 for Logyard or appstore.libevent-dev and libffi-devstackato clientOFFLINE rather than STAGINGkato relocate commandskato node setup micro --no-start silently failskato always returns 0 exit code due to passing output through teekato data import of app fails and halts importkato node reset factory leaves VMs in a corrupt statekato relocate containers failing - Device or resource busykato node rename hangs waiting for passwordkato restart on core node causes issues on other nodeskato op max_client_upload fails while restarting cc_nginx - undefined method `join' for "cc_nginx":String (NoMethodError)kato node reset soft has been removedkato process ready --block does not actually timeout as expectedcmd.rb files for kato commands with more than three sub-commandskato op set_timezone not updated to use dockerkato data import ignores manifest.ymlkato reportkato log tail drain should use UDP instead of TCPkato patch when fetching patch manifest from get.stackato.comkato patch status --json errorkato data import does not always detect worker apps without a URLkato node rename crashing--upgrade option to kato op import_from_yaml_fileskato status hangs and stacktraces when a node goes offlinekato node retire -n <node> hangs, apparently waiting for a sudo passwordkato report broken after 3.2.1 -> 3.4.1 upgradekato inspect returns 'Docker configuration is not valid Cannot locate docker image :latest'kato node upgrade --skip-confirmation does not skip confirmationkato node upgrade remote upgrades not re-throwing exceptions after creating upgrade-failed flat file.stackato logs using websocketskato log tail and app log streamskato op regenerate ...prevent_x_spoofing router2g configuration with load balancersdispatchWebSocketErr function and terminate socket to allow long polling fallback for the clients--health-timeout and stackato.yml/manifest.yml supportvendor_version information of /info endpointkato data import/export cannot clearly delete the previous first admin account#<CF::UAA::InvalidTokenFormat: Not enough or too many segmentsmssql service is not created during IronFoundry setup scriptstackato tunnel fails against 3.2 servertarget on Windowsupdate-service-broker should offer to change URL, username & password attributesmarketplace commandstackato deletetrace outputs garbage charactershttps://stackato buildpacks display bugstackato help pushstackato user-info is broken for non admin user--space option is strictly enforced with --no-prompt but not withoutlogs -f alias for logs --followstackato curlstackato services fails against 3.0.1--droplets option for quota create and quota configurestackato logs --tail does not stream the logsstackato tunnel fails with error - got keep-zip, expected parameter namestackato push --as ... does not work without stackato.ymldelete-service as adminservice --json / credentials field missing from services outputstackato pushhistory command--code-only option to stackato rollbackpush --pathopen should deduce the app name from the directory namefiles gives "~" rather than "~/app"service-plans and help service-plan outputcreate-buildpack - support directory as source of the BP (auto-create zip for upload)kato node upgrade to handle interruptionskato node reset now resets services datakato op regenerate ssh-keysstackato/stack/alsek becomes stackato/stack-alsekkatocd node-env to Clone Repo commandskato node reset failing to remove Docker containers/imageskato patch now respects proxy settingskato op remap_hosts fails after importkato node reset does not delete serviceskato node reset soft ends with bash syntax errorkato node reset does not change the setting to defaultkato role add rabbit fails to associate with CC (403)kato node remove hangs for 2-3 minutes if node is unreachablekato patch update now updates on all nodes by default(--local and --node to override)kato status crashes when an external node initially becomes unavailablekato op upstream_proxy set should prompt to restart DEA, not Stagerkato node remove a batch operationkato sub-commandskato attach crashes if a unreachable IP address is givenkato export prompts for password when exporting PostgreSQL servicekato patch cluster installexpress.bodyParser and use Connect.json insteadmax_container_processes) from 50 to 256stackato crashesstackato scp docs missing some helpful contentstackato apps --all does not workstackato scp should behave more like real scpstackato scp appname internal errorstackato help admin too verbose at top level--token-file does not create empty token filestackato open tries to open http:/// when the app has no URLstackato open command help clarificationstackato delete logs debug datastackato link-user-org should not have --developerstackato info must show that (v2) is the API version-group for Stackato 2.x targetsError: cannot read "mymap(r1558)": no such element in arraymax_user_drains quota replaced by max_drains_per_appkato report and kato cluster --all (-a) options changed to --cluster (-c)kato node resetkato process stop config_redis (and sup stop config_redis) hangkato report no longer asks for sudo passwordlibpq-dev and bundler available in the container--env settings are always written--url to appstore push APIDATABASE_URL and POSTGRESQL_URL environment variables now use postgres:// instead of postgresql://JDBC_DATABASE_URL environment variable is now available (jdbc\:postgresql\://)--timeout for start and push commands--name alias for --asstackato run --all option(un)mappre-push hooks-d debug port handlingswitch commands against known orgs/spacesstackato tunnel with CF v1 targetskato config under fence max_container_processeskato config under stager app_memory_multiplierkato node retire to gracefully shut down a DEA, moving its application instances on other available nodes first.kato admin ... commands removed; run the corresponding stackato admin ... commands or the web consolekato node upgrade command to support node-level upgradeskato process stop fence now correctly stops fence processkato node attach now checks for version compatibility when adding a node to a cluster--no-stop option removed from kato role remove command.kato config no longer supports node-specific config; --node option removed.kato node setup firstuser to require default organizationNOKOGIRI_USE_SYSTEM_LIBRARIES=true).kato log drain status command for showing drain statuskato config)read_limit setting for apptail (default 16MB) to cope with extremely large application log filesrouter2g-access.log)kato role add rabbit3host subcommandpyopenssl is now installable via PyPM without an ActivePython Business Edition licenseArgumentError in get_datastore (1 for 2)/home/stackato/.ssh/known_hostscloud_controller X-Accel-Mapping header is missingkato status role reporting overlapkato op upstream proxy ... now correctly modifies the LXC templatekato op regenerate postgresqlFATAL -- Cannot create DEA pid file: Process already runningparse_info!': undefined method `[]')stackato update preserves
previously set environment variables. New --env-mode replace
option enables the old behavior of removing or resetting environment
variables.token command for interactive token-based
authentication with external SSO systems.kato log tailstackato deleteApp::cpanminus supports installation via URLs and from gitstdsync usagekato users has moved to kato data userskato node list to list all nodes and which roles
are running on themkato tail skips logyard INFO records unless --raw is passed.kato node rename to handle existing pushed apps when using mdnskato data export exception if memcached services was never usedstackato dbshell supportstackato files logs/staging.log not returning
staging.logwait_for_fs.rb (when filesystem service is used)lxctrl processes (less memory use)kato start|stop|restart now manipulates roles, not processes.kato enable|disable is now kato role add|removekato attach is now kato node attachkato setup micro is now kato node setup micro, kato setup core is now kato node setup core and kato setup load_balancer is now kato node setup load_balancerkato setup controller and kato setup node have been retired. Use kato node setup core and kato node attach instead.kato drain add options are passed. Dropped --option key=value --option key=value in favor of a just key=value key=valuekato role add|remove (previously kato enable|disable), --all-but and --only are now boolean flags that act on the list of roles given. Therefore kato enable --all-but dea,prealloc becomes kato role add --all-but dea preallockato role remove no longer supports the flag --no-stopkato rename is now kato node renamekato role list to list all roles and number of nodes they are enabled on.kato config now takes action keywords set, get, del, push and pop. This is not backwards compatible. Example is kato config set <component> <key-path> <value>.kato tail is now kato log tailkato drain is now kato log drainkato process start|stop|restart can now be used to manage processes. However, managing roles via kato start|stop|restart is recommended.kato ready, for checking the ready status of processes, is now kato process ready--node to kato process ready to check status or process on any node.kato config default output is YAMLkato debug lsdoozer has been removed. Use kato config get --flat instead.kato status functionality that showed just a list of processes and their statuses can now be found under kato process listkato import is now kato data import.kato export is now kato data export.file:/// drain for debugging purposes.kato commands and command naming:kato import/export uses the cluster-aware back end formerly accessed
with kato migrate (deprecated)kato history command shows a history of administrative kato
commands ran across the cluster.kato ls in favor of kato status --allcontrib plugin frameworks directoryCannot connect to doozerd warning in various cluster situations.stackato drains add (user-level)stackato logs runs much faster, returns logs from all instances, and returns logs from stopped/crashed instancesstackato logs --follow tails logs from all instanceskato migrate - next generation cluster live migration toolkato op static_ip - intuitive tool for static-ip configurationkato relocate ... - move key shared data to a new
location (e.g. persistent storage).csv format)kato setup (BETA - not for production use)scp to move data to/from app instancespypmframework: buildpack in stackato.yml for ruby apps, BUILDPACK_URL can be omitted.copy_source_files for non java_* frameworksautobecomestackato run dbexport exports mysql and psql databasesVCAP_SERVICES service key to use an unversioned service name
(version remains in label key) to bring parity to client service creationuniq keys in kato importaspdotnet [95122] framework auto-detectionprealloc queue lengthjava_ee framework via Apache TomEE--use-mirrors) when installing
packagesstackato push|update will now tail the staging.log file simultaneously
(like Heroku).stackato dbshell is deprecated in favor of stackato run
dbshell. The later works on cluster environment, and does not
require local database clients to be installed.STACKATO_SERVICES as an alternative to VCAP_SERVICES; PORT as
an alias for VCAP_APP_PORT.DATABASE_URL, MYSQL_URL, POSTGRESQL_URL, REDIS_URL,
MONGODB_URL, RABBITMQ_URLstackato
update--stackato-debug <host>:<port> enables
Django, Rails, and Sinatra get requests to be debugged in DBGP-supporting
apps like Komodo IDE with a STACKATO_DEBUG=1 parameter, and enables PHP
debugging with the XDEBUG_SESSION_START=1 and XDEBUG_SESSION_STOP=1
parameters.lein deps to download dependenciesvmc) manifest.ymlstackato-admin report: wrap logs and other system information
for future diagnosis of a problem.$app_http_proxy is now available to specify a application specific $http_proxystackato-admin upstream-proxy command helps setting this variable,
check the docs or stackato-admin help for more information.vcap logs in /tmp/vcap-run/ to $HOME/stackato/logs/stackato-admin become <node> now allows renaming operations
on the VM, allowing you to specify the FQDN and local hostname.all role has also been added to stackato-admin become to assign
all the available roles.PYTHONPATH to contain correct value in the "run" commandudev networking rules have been disabledssh connections to to application instances (experimental)stackato run based on SSH.$STACKATO_UWSGI environment
variable.$PROCESSES_WEB environment variable (TBD for other frameworks)$STACKATO_UWSGI environment
variable.$PROCESSES_WEB environment variable (TBD for other frameworks)$PATH before
starting the applicationstackato-admin start
dashboard)$PYTHONPATH--catch-exceptions by default:port back to VCAP_APPLICATION for the cloudfoundry runtime to work./include dir. (for example, greenlet)staging.loghttp://docs.stackato.localstackato runstackato envdbshell command to invoke database clientsstackato run:$PATH; add all runtimes to $PATHnpm installPython
dependencies
Python 3.x support for 'Generic Python' framework
Perl
app.psgi fileperl514 runtime nameMakefile.PLlogs/staging.log and logs/ppm4.logRuby:
stackato run:
Put runtime's path in front of $PATH
Close STDIN descriptor to prevent hanging processes
Support --timeout option (default = 1 minute)
GEM_{HOME,PATH} environment variables so dependentbinaries (for example, rake) can be run
Remove the vcap/installer/remote after an install
stackato-admin uses yaml parser instead of grep
Install script is now called install.sh and accepts parameters
Do not regenerate the mysql password on restarting ./install
Do not pull uWSGI directly from mercurial repo
Allow git branch argument for ./install