Commit graph

715 commits

Author SHA1 Message Date
Michael Schurter 4b83eba599 Use seen more conservatively 2017-07-24 16:48:40 -07:00
Michael Schurter cdf138eb27 Always increment failures...
...as it's used in calculating the backoff
2017-07-24 15:37:53 -07:00
Michael Schurter 809724ad8d Track whether Consul has ever been seen
Need a way to squelch Consul operation errors on shutdown. If it's never
been seen don't log errors about deregs failing.
2017-07-24 12:12:02 -07:00
Michael Schurter edbe62a879 Synchronously deregister agent on shutdown
Fixes #2891

Previously the agent services and checks were being asynchrously
deregistered on shutdown, so it was a race between the sync goroutine
deregistering them and Nomad shutting down.

This switches to synchronously deregister agent serivces and checks
which doesn't really have a downside since the sync goroutines retry
behavior doesn't help on shutdown anyway.
2017-07-24 11:40:37 -07:00
Alex Dadgar 553bc91725 Parallel client tests (#2890)
* alloc_runner

* Random tests

* parallel task_runner and no exec compatible check

* Parallel client

* Fail fast and use random ports

* Fix docker port mapping

* Make concurrent pull less timing dependant

* up parallel

* Fixes

* don't build chroots in parallel on travis

* Reduce parallelism on travis with lxc/rkt

* make java test app not run forever

* drop parallelism a little

* use docker ports that are out of the os's ephemeral port range

* Limit even more on travis

* rkt deadline
2017-07-22 19:04:36 -07:00
Alex Dadgar 4dd5d943c7 remove root requirement on consul integration check 2017-07-21 19:32:41 -07:00
Alex Dadgar 56f9cf86df Speed up client startup 2017-07-20 22:34:24 -07:00
Alex Dadgar c106df9215 Switch to in-process agent 2017-07-20 21:07:32 -07:00
Alex Dadgar d019eed363 Merge pull request #2874 from hashicorp/f-command-agent-tests
Parallelize the command/agent tests and add new test agent
2017-07-20 20:27:49 -07:00
Alex Dadgar 5df9be0ccb Fix bootstrapping and waiting 2017-07-20 20:15:37 -07:00
Alex Dadgar 4e90d56098 More parallel 2017-07-20 09:36:34 -07:00
Alex Dadgar 9037693436 New test agent 2017-07-19 22:14:36 -07:00
Alex Dadgar 9a2a5af608 Don't print atlas 2017-07-19 20:25:06 -07:00
Alex Dadgar 18c7043a30 Merge pull request #2866 from hashicorp/f-autocomplete-agent
Agent command autocompletes to hcl/json files
2017-07-19 13:18:32 -07:00
Michael Schurter c9e4c041b3 Too lazy to remember the right formatter for floats 2017-07-19 11:53:18 -07:00
Alex Dadgar b4b50b636f Fix predictor 2017-07-19 11:51:01 -07:00
Alex Dadgar 871cdcb932 Agent command autocompletes to hcl/json files 2017-07-19 11:28:16 -07:00
Michael Schurter 40c2d4e5eb Merge pull request #2858 from hashicorp/b-2849-deploy-json
Implement -json for job deployments
2017-07-19 10:15:01 -07:00
Michael Schurter 125a3fb2f9 Error -> Errof 2017-07-19 10:00:57 -07:00
Alex Dadgar 747d67eb3f Allow tuning of heartbeat ttls
This PR allows tuning of heartbeat TTLs. An example of very aggressive
settings is as follows:

```
server {
  heartbeat_grace = "1s"
  min_heartbeat_ttl = "1s"
  max_heartbeats_per_second = 200.0
}
```
2017-07-19 09:38:35 -07:00
Michael Schurter 99d1486f32 Never remove unknown agent services
Fixes #2827

This is a tradeoff. The pro is that you can run separate client and
server agents on the same node and advertise both. The con is that if a
Nomad agent crashes and isn't restarted on that node in the same mode
its entry will not be cleaned up.

That con scenario seems far less likely to occur than the scenario on
the pro side, and even if we do leak an agent entry the checks will be
failing so nothing should attempt to use it.
2017-07-18 13:23:01 -07:00
Alex Dadgar 45712c6ca3 test fixes 2017-07-07 14:11:27 -07:00
Alex Dadgar bf2dafb8e9 check id method name changed 2017-07-07 12:15:09 -07:00
Alex Dadgar 1cb877699a Disallow update stanza on batch jobs
This PR:
* disallows update stanzas on batch jobs
* undeprecates the stagger field
* changes the way warnings are returned
2017-07-07 12:11:39 -07:00
Alex Dadgar 5457bb7962 Job stability 2017-07-07 12:10:04 -07:00
Alex Dadgar 09dfa2fc10 Rename CreateDeployments and remove cancelling behavior in state_store 2017-07-07 12:10:04 -07:00
Alex Dadgar 067ed86a47 Client watches for allocation health using task state and Consul checks
This PR adds watching of allocation health at the client. The client can
watch for health based on the tasks running on time and also based on
the consul checks passing.
2017-07-07 12:10:04 -07:00
Alex Dadgar abf34204cc JobVersions returns struct with optional diff 2017-07-07 12:05:57 -07:00
Alex Dadgar c643e6b0d1 Add config options 2017-07-07 12:05:56 -07:00
Alex Dadgar f233629a4f job deployment endpoint + api 2017-07-07 12:05:56 -07:00
Alex Dadgar 580eed5c88 HTTP Endpoints 2017-07-07 12:03:11 -07:00
Michael Schurter cab28b2963 Fix api endpoint test 2017-07-06 10:45:44 -07:00
Michael Schurter 0d3bdf7210 Add support for go-getter modes
Fixes #2678
2017-07-06 10:45:44 -07:00
Michael Schurter b9c9e6e557 Fix no_host_uuid parsing
Need to pointerify it to default to true since we can't tell false from
unset if it's not a pointer.
2017-07-03 17:41:20 -07:00
Michael Schurter 6e7cc3964e Merge pull request #2709 from hashicorp/f-advertise-docker-ips
Advertise driver-specific addresses
2017-07-03 14:04:12 -07:00
Michael Schurter d9e032aabf Merge pull request #2735 from hashicorp/f-no_host_uuid-true
Default no_host_uuid to true instead of false
2017-07-03 13:18:25 -07:00
Alexandre Dantas 100b51ac6a Fixing issue where use_node_name was always been set as false when merging telemetry configurations 2017-07-02 00:31:09 -03:00
Michael Schurter e9c357187c Properly normalize IPv6 addresses
A fix to #2739 instead of forcing IPv6 users to always specify a port as
well.

Prior to this commit IPv6 advertise addresses which lacked a port would
fail instead of having the default port added because
`net.SplitHostPort(someipv6)` returns a different error than
`net.SplitHostPort(someipv4)`.
2017-06-29 10:46:31 -07:00
Michael Schurter a863ead30e Fix test error formats 2017-06-26 12:53:43 -07:00
Michael Schurter e81252ba45 Default no_host_uuid to true instead of false
The host UUID isn't unique in many virtualized cases and of dubious
value even when it is univerally unique. Default to a random UUID.
2017-06-23 16:23:01 -07:00
Michael Schurter 5b59bea67b Move caonicalization from nomad/structs/ to api/ 2017-06-21 17:19:08 -07:00
Michael Schurter 9da78ae25f Remove debug logging 2017-06-21 17:19:08 -07:00
Michael Schurter c0eff81383 Fix Service.AddressMode changes during task updates 2017-06-21 17:19:08 -07:00
Michael Schurter 67d154a274 Test driver network advertisement and checks 2017-06-21 17:19:08 -07:00
Michael Schurter b9bfb84b53 Implement DriverNetwork and Service.AddressMode
Ideally DriverNetwork would be fully populated in Driver.Prestart, but
Docker doesn't assign the container's IP until you start the container.

However, it's important to setup the port env vars before calling
Driver.Start, so Prestart should populate that.
2017-06-21 17:19:08 -07:00
Michael Schurter 95a00cbef1 Fix path used by Nomad Server HTTP Check
Fixes #2701
2017-06-21 10:41:28 -07:00
Michael Schurter ffc2b36dc7 Merge pull request #2636 from hashicorp/f-gc-alloc-limit
Add new gc_max_allocs tuneable
2017-05-30 16:14:09 -07:00
Michael Schurter dd51aa1cb9 Merge pull request #2654 from hashicorp/f-env-consul
Add envconsul-like support and refactor environment handling
2017-05-30 14:40:14 -07:00
Michael Schurter a7e26e0a3e Don't autoadvertise private ip if bind=localhost
A slight improvement to #2399 - if bind is localhost, return an error
instead of advertising a private ip. The advertised ip isn't valid and
will just cause errors on use. It's better to fail with an error message
instructing users how to fix the problem.
2017-05-30 11:47:29 -07:00
Michael Schurter bbf299dde1 Fix config parsing test
Went overboard before I realized there's only one test case.
2017-05-30 11:39:26 -07:00