Tom Clegg [Fri, 2 Oct 2015 23:54:36 +0000 (19:54 -0400)]
7410: Add keepexercise
Tom Clegg [Tue, 13 Oct 2015 15:17:49 +0000 (11:17 -0400)]
7159: Fix error handling when reading full size block. refs #7159
Tom Clegg [Mon, 12 Oct 2015 16:40:38 +0000 (12:40 -0400)]
Warn about unhandled case if broken node has no ping time. refs #7286
Tom Clegg [Mon, 12 Oct 2015 16:06:47 +0000 (12:06 -0400)]
Merge branch '7159-empty-blob-race' refs #7159
Tom Clegg [Mon, 12 Oct 2015 13:33:24 +0000 (09:33 -0400)]
7159: Shorten race waits during generic tests
Tom Clegg [Fri, 9 Oct 2015 21:09:42 +0000 (17:09 -0400)]
7159: Log when waiting for get/put races
Tom Clegg [Thu, 8 Oct 2015 18:00:00 +0000 (14:00 -0400)]
7159: Exclude new empty blocks from index.
Tom Clegg [Thu, 8 Oct 2015 17:30:18 +0000 (13:30 -0400)]
7159: Test race deadline
Tom Clegg [Thu, 8 Oct 2015 16:52:17 +0000 (12:52 -0400)]
7159: Work around CreateBlob race by polling for updates when a brand new blob is found empty.
Tom Clegg [Fri, 9 Oct 2015 19:48:28 +0000 (15:48 -0400)]
Merge branch '7491-keepclient-bugs' refs #7491
Tom Clegg [Fri, 9 Oct 2015 18:51:13 +0000 (14:51 -0400)]
Merge branch '7167-blob-sign-sdk' refs #7167
Tom Clegg [Fri, 9 Oct 2015 18:28:29 +0000 (14:28 -0400)]
7167: Deobfuscate variable names
Tom Clegg [Fri, 9 Oct 2015 18:20:15 +0000 (14:20 -0400)]
7167: Update tests and comments to new error vars.
Tom Clegg [Thu, 8 Oct 2015 21:50:22 +0000 (17:50 -0400)]
7167: Fix up comments
Tom Clegg [Thu, 8 Oct 2015 21:33:55 +0000 (17:33 -0400)]
7167: Replace duplicate tests with PermissionSecret tests
Brett Smith [Fri, 9 Oct 2015 15:24:50 +0000 (11:24 -0400)]
Version the CWL runner's dependency on cwltool.
cwltool development is continuing on with API-incompatible changes.
No issue #.
Tom Clegg [Thu, 8 Oct 2015 21:17:46 +0000 (17:17 -0400)]
7167: Tidy up errors. Remove extra comment copy.
radhika [Wed, 7 Oct 2015 20:47:56 +0000 (16:47 -0400)]
7167: move perms code from keepstore into keepclient go SDK.
Tom Clegg [Thu, 8 Oct 2015 20:20:46 +0000 (16:20 -0400)]
6967: Update test to match improved code.
refs #6967
Tom Clegg [Tue, 6 Oct 2015 17:48:05 +0000 (13:48 -0400)]
7491: Ensure status channel stays open until all upload workers finish.
Tom Clegg [Tue, 6 Oct 2015 17:39:12 +0000 (13:39 -0400)]
7491: Fix error handling/reporting in keepclient/GET
Tom Clegg [Thu, 8 Oct 2015 19:24:19 +0000 (15:24 -0400)]
Merge branch '6967-yaml-format' closes #6967
Tom Clegg [Thu, 8 Oct 2015 18:46:22 +0000 (14:46 -0400)]
6967: More helpful comment & assertion failure message
Tom Clegg [Thu, 8 Oct 2015 18:45:33 +0000 (14:45 -0400)]
6967: Use git status --porcelain to isolate from user config
Tom Clegg [Wed, 7 Oct 2015 15:11:14 +0000 (11:11 -0400)]
6967: Move source_version detection code from config yaml to lib/app_version.rb.
Tom Clegg [Wed, 7 Oct 2015 14:00:16 +0000 (10:00 -0400)]
6967: Move source_version detection code from config yaml to lib/app_version.rb.
Tom Clegg [Wed, 7 Oct 2015 14:01:22 +0000 (10:01 -0400)]
6967: Treat blob_signing_key like a secret in `rake config:check`.
Peter Amstutz [Wed, 7 Oct 2015 18:09:40 +0000 (14:09 -0400)]
Merge branch '6142-cancel-slurm' closes #6142
Tom Clegg [Wed, 7 Oct 2015 15:44:25 +0000 (11:44 -0400)]
Merge branch '7254-dont-lose-replication-arg' closes #7254
Brett Smith [Wed, 7 Oct 2015 14:54:08 +0000 (10:54 -0400)]
Merge branch '7435-node-manager-shutdown-cleanup-wip'
Closes #7435, #7445.
Brett Smith [Wed, 7 Oct 2015 14:47:23 +0000 (10:47 -0400)]
7254: Test arv-put preserves replication when cache load fails.
Peter Amstutz [Wed, 7 Oct 2015 14:38:46 +0000 (10:38 -0400)]
6142: Only resume from 'drng' or 'drain'. Add/fix tests.
Brett Smith [Fri, 2 Oct 2015 15:07:27 +0000 (11:07 -0400)]
7435: Node Manager stops trying to shut down delisted cloud nodes.
If the underlying node is gone, trying to destroy it in the cloud will
almost certainly fail. It's hard to predict what will happen to
related actions like draining the node in SLURM. Just cancel the
attempt, and trust other systems like SLURM and Crunch to deal with
the disappearance on their own.
Tom Clegg [Tue, 6 Oct 2015 21:10:28 +0000 (17:10 -0400)]
7254: Test that replication arg is passed through to KeepClient.put()
Peter Amstutz [Tue, 6 Oct 2015 20:42:51 +0000 (16:42 -0400)]
6142: If self._set_node_state('RESUME') in cancel_shutdown() returns non-zero,
check the node state and only retry if the node is in 'drain' or 'draining'.
Tom Clegg [Tue, 6 Oct 2015 19:04:13 +0000 (15:04 -0400)]
7254: Do not forget -replication arg when failing to load resume state.
Peter Amstutz [Tue, 6 Oct 2015 13:17:30 +0000 (09:17 -0400)]
Merge branch '7286-nodeman-destroy-broken-nodes' closes #7286
Peter Amstutz [Tue, 6 Oct 2015 01:33:00 +0000 (21:33 -0400)]
7286: Add comments clarifying arvados_node_missing() and broken(). Also bump
up version dependency to dev4.
Tom Clegg [Fri, 2 Oct 2015 22:09:39 +0000 (18:09 -0400)]
7214: Fix "X-Keep-Replicas-Stored: 0" header when block is already present. refs #7214
Tom Clegg [Fri, 2 Oct 2015 20:06:12 +0000 (16:06 -0400)]
Merge branch '7241-azure-blob-volume' closes #7241
Peter Amstutz [Thu, 1 Oct 2015 17:00:00 +0000 (13:00 -0400)]
7286: Add BaseHTTPError to list of "cloud errors"
Peter Amstutz [Thu, 1 Oct 2015 13:32:33 +0000 (09:32 -0400)]
7286: Add drain* and fail* to SLURM_END_STATES, because the '*' means the node
is out of contact with slurm.
Peter Amstutz [Wed, 30 Sep 2015 21:16:09 +0000 (17:16 -0400)]
7286: Fix double count of missing nodes in shutdown
Peter Amstutz [Wed, 30 Sep 2015 20:26:46 +0000 (16:26 -0400)]
7286: Missing nodes are considered in "excess" count (reverts previous change). Added test. Also remove debug log statement.
Peter Amstutz [Wed, 30 Sep 2015 18:23:25 +0000 (14:23 -0400)]
7286: Compute "missing" based on "last_ping_at" instead of using API server's
buggy "status" field.
radhika [Wed, 30 Sep 2015 15:53:20 +0000 (11:53 -0400)]
closes #7200
Merge branch '7200-keepproxy-index-api'
radhika [Wed, 30 Sep 2015 15:46:41 +0000 (11:46 -0400)]
7200: more compact CheckAuthorizationHeader block
radhika [Wed, 30 Sep 2015 15:28:36 +0000 (11:28 -0400)]
7200: Use io.Copy instead of reading all bytes and writing to response. Much improved keep proxy test with code reuse.
Peter Amstutz [Wed, 30 Sep 2015 14:35:04 +0000 (10:35 -0400)]
7286: Add test that "missing" nodes are not counted towards "busy" (but are
counted towards node max).
Brett Smith [Wed, 30 Sep 2015 14:18:18 +0000 (10:18 -0400)]
7207: Remove `arv keep check` and `arv keep less`.
The implementations are no longer maintained and these are stale
references. Closes #7207.
Brett Smith [Wed, 30 Sep 2015 13:10:27 +0000 (09:10 -0400)]
7263: crunch-job checks for refreshes every two seconds.
This avoids the possibility that a constant stream of data from tasks
can prevent the job from being canceled. Refs #7263.
radhika [Wed, 30 Sep 2015 02:19:42 +0000 (22:19 -0400)]
7200: no need to set content length header in keepproxy IndexHandler since the response is always terminated by a blank new line.
radhika [Wed, 30 Sep 2015 00:31:00 +0000 (20:31 -0400)]
7200: Use if statement instead of switch to check http method in keepclient; strip terminating empty line from response in keepclient.
Update tests accordingly. Also, improve keepproxy test to verify that "other" locators are included when no prefix is used in GetIndex requests.
radhika [Tue, 29 Sep 2015 21:44:26 +0000 (17:44 -0400)]
Merge branch 'master' into 7200-keepproxy-index-api
Peter Amstutz [Tue, 29 Sep 2015 20:43:41 +0000 (16:43 -0400)]
7286: Tests for new "missing and broken" shutdown policy.
Tom Clegg [Tue, 29 Sep 2015 18:22:22 +0000 (14:22 -0400)]
Merge branch 'master' into 7241-azure-blob-volume
Conflicts:
services/keepstore/volume_generic_test.go
services/keepstore/handlers_with_generic_volume_test.go
Tom Clegg [Tue, 29 Sep 2015 18:19:40 +0000 (14:19 -0400)]
7241: Use If-Match header to address Delete/Put races.
Ward Vandewege [Tue, 29 Sep 2015 18:03:52 +0000 (14:03 -0400)]
Merge branch '7330-improved-sso-package'
closes #7330
Ward Vandewege [Tue, 29 Sep 2015 17:50:44 +0000 (13:50 -0400)]
Merge branch 'master' into 7330-improved-sso-package
refs #7330
Ward Vandewege [Tue, 29 Sep 2015 17:48:19 +0000 (13:48 -0400)]
More language tweaks.
refs #7330
radhika [Tue, 29 Sep 2015 16:13:04 +0000 (12:13 -0400)]
refs #6859
Added include statement to the script.
Nico Cesar [Tue, 29 Sep 2015 14:28:05 +0000 (10:28 -0400)]
7330: 'edit' should be replaced by 'editor' that will go to the default editor of the ditro
refs #7330
Peter Amstutz [Tue, 29 Sep 2015 14:27:10 +0000 (10:27 -0400)]
7286: Move logic to shut down newly booted nodes nodes that haven't pinged to
ComputeNodeMonitorActor. Shut down nodes if they have "missing" status and are
"broken" according to the cloud client. Don't count "missing" nodes as "up"
when deciding whether to boot more nodes.
Nico Cesar [Tue, 29 Sep 2015 13:46:50 +0000 (09:46 -0400)]
7330: making sso install guide similar to workbench and api server
refs #7330
Nico Cesar [Tue, 29 Sep 2015 13:28:21 +0000 (09:28 -0400)]
7330: apt-get command had a typo
refs #7330
Nico Cesar [Tue, 29 Sep 2015 13:02:06 +0000 (09:02 -0400)]
7330: curl isnt available by default in ubuntu 14.04. added the apt-get
refs #7330
Ward Vandewege [Tue, 29 Sep 2015 00:17:12 +0000 (20:17 -0400)]
Another tweak for the SSO server installation guide: the PostgreSQL
service needs to be started manually if it was pulled in as a dependency
by arvados-sso-server.
refs #7330
Ward Vandewege [Tue, 29 Sep 2015 00:09:00 +0000 (20:09 -0400)]
More improvements for the SSO server installation guide.
refs #7330
Ward Vandewege [Mon, 28 Sep 2015 21:02:48 +0000 (17:02 -0400)]
Update the SSO installation documentation to reflect the new and
improved packages.
refs #7330
radhika [Mon, 28 Sep 2015 19:56:32 +0000 (15:56 -0400)]
Merge branch 'master' into 7200-keepproxy-index-api
radhika [Mon, 28 Sep 2015 19:55:43 +0000 (15:55 -0400)]
7200: rename error variables to meet golint suggestions.
radhika [Mon, 28 Sep 2015 19:28:24 +0000 (15:28 -0400)]
7200: The incomplete response when no such prefix exists will be "\n". Update keepclient and keepproxy to expect this.
Also, added some more comments and golint checks for the newly added code.
Nico Cesar [Mon, 28 Sep 2015 18:54:33 +0000 (14:54 -0400)]
6034: added scafolding of the first initscript
refs #6034
Tom Clegg [Mon, 28 Sep 2015 18:41:24 +0000 (14:41 -0400)]
7241: Use new CreateBlockBlobFromReader and SetBlobMetadata APIs for Put and Touch.
Ward Vandewege [Mon, 28 Sep 2015 14:29:56 +0000 (10:29 -0400)]
Ruby 2.1.7 is out, update documentation accordingly.
No issue #
radhika [Mon, 28 Sep 2015 14:10:48 +0000 (10:10 -0400)]
7200: add GetIndex function to keepclient; add IndexHandler to keepproxy.
Brett Smith [Mon, 28 Sep 2015 13:53:20 +0000 (09:53 -0400)]
Merge branch 'pr/29'
Closes #7388.
LunarShaddow [Mon, 28 Sep 2015 02:48:03 +0000 (10:48 +0800)]
Change ssh public key to 700 after key-sync
Tom Clegg [Fri, 25 Sep 2015 19:31:27 +0000 (15:31 -0400)]
7241: Do not bother trying to add a volume if credential args are missing/empty.
Tom Clegg [Fri, 25 Sep 2015 18:56:27 +0000 (14:56 -0400)]
7241: Add -azure-storage-replication flag.
Tom Clegg [Thu, 24 Sep 2015 22:31:43 +0000 (18:31 -0400)]
7241: Stub Azure API calls
radhika [Fri, 25 Sep 2015 17:27:24 +0000 (13:27 -0400)]
closes #7161
Merge branch '7161-support-all-service-types'
radhika [Fri, 25 Sep 2015 16:44:15 +0000 (12:44 -0400)]
Merge branch '7161-support-all-service-types' of git.curoverse.com:arvados into 7161-support-all-service-types
radhika [Fri, 25 Sep 2015 14:13:49 +0000 (10:13 -0400)]
7161: Use default value of 1 for max_replicas_per_service and set it to None when writable non-disk type services are found.
radhika [Thu, 24 Sep 2015 18:57:40 +0000 (14:57 -0400)]
7161: white space
radhika [Thu, 24 Sep 2015 18:52:39 +0000 (14:52 -0400)]
7161: clarify max_replicas_per_service
radhika [Wed, 23 Sep 2015 19:32:44 +0000 (15:32 -0400)]
7161: Look for all service hints, not just keep proxy. Also, use thread count of 1 when it not a disk.
radhika [Fri, 25 Sep 2015 16:15:36 +0000 (12:15 -0400)]
closes #7162
Merge branch '7162-support-service-types'
radhika [Fri, 25 Sep 2015 16:13:25 +0000 (12:13 -0400)]
7162: Removed the call to get keep_disks when there is an error getting keep_services,
since keep_services has been available since May 2014, #2776 according to Tom.
radhika [Fri, 25 Sep 2015 16:05:44 +0000 (12:05 -0400)]
Merge branch 'master' into 7162-support-service-types
Brett Smith [Fri, 25 Sep 2015 14:46:00 +0000 (10:46 -0400)]
Merge branch '7227-crunch-job-stricter-unmount-wip'
Closes #7227, #7376.
Brett Smith [Wed, 23 Sep 2015 17:16:41 +0000 (13:16 -0400)]
7227: crunch-job only unmounts Keep mounts under $CRUNCH_TMP.
This prevents it from trying to unmount other Keep mounts elsewhere on
the system, and considering the cleanup step failed because of that.
Brett Smith [Fri, 25 Sep 2015 14:38:37 +0000 (10:38 -0400)]
Merge branch '6967-application-yml-without-git-wip'
Closes #6967, #7375.
Brett Smith [Wed, 23 Sep 2015 15:38:06 +0000 (11:38 -0400)]
6967: Extend application.default.yml source_version detection.
* In production, get source_version from our package's
git-commit.version, assuming it's available.
* When we get source information from Git, suppress errors from the
user, and only use the information if all queries succeed.
Brett Smith [Wed, 23 Sep 2015 15:34:53 +0000 (11:34 -0400)]
6967: Remove trailing whitespace in Workbench's application.default.yml.
radhika [Fri, 25 Sep 2015 14:13:49 +0000 (10:13 -0400)]
7161: working
radhika [Fri, 25 Sep 2015 12:32:46 +0000 (08:32 -0400)]
Merge branch 'master' into 7161-support-all-service-types
radhika [Fri, 25 Sep 2015 12:29:59 +0000 (08:29 -0400)]
Merge branch 'master' into 7162-support-service-types
radhika [Fri, 25 Sep 2015 12:28:04 +0000 (08:28 -0400)]
7162: updated comment
Peter Amstutz [Thu, 24 Sep 2015 20:36:59 +0000 (16:36 -0400)]
Merge branch '7181-crunch-missing-git' closes #7181
projects / arvados.git / log