Commit Graph

358 Commits (8901125a5a68ff280e790f6320d63c6d5cc7cdcd)

Author SHA1 Message Date
Philip O'Toole 878663e1e3 Periodic upload of stats to Enterprise 2015-10-19 15:25:07 -07:00
Philip O'Toole 667ad3342a Refactor registration as a service
Registration also involves statistics and diagnostics upload, for the
purposes of remote management. This means there will be long-running
goroutines in effect. Therefore move the code to a service model.
2015-10-19 15:01:14 -07:00
Philip O'Toole ef72c3c64d Fix typo in retention service comment
[ci skip]
2015-10-19 14:24:25 -07:00
Philip O'Toole ff18bf7213 Make Open() and Close() on Graphite sync'ed
This will ensure that these operations don't run concurrently. This
change also ensures nil batchers are not closed.

Fixes issue #4494.
2015-10-19 11:13:31 -07:00
David Norton e73a8e423c fix #4472:too many points in the GROUP BY interval 2015-10-16 07:17:14 -04:00
Nathaniel Cook cb1aaa8e42 Merge pull request #4375 from influxdb/subscriptions
Feature add subscriber service for creating/dropping subscriptions
2015-10-15 09:17:26 -06:00
Philip O'Toole 485c446e98 Correct typos in UDP README
[ci skip]
2015-10-15 07:48:34 -07:00
Sean Beckett 82f104a8b1 Merge pull request #4436 from influxdb/tag-names-to-keys
WIP tag name --> tag key, field name --> field key
2015-10-14 16:02:46 -07:00
Nathaniel Cook 8b31007aa7 Adds subscriber service for creating/dropping subscriptions to the
InfluxDB data stream.
2015-10-14 15:23:45 -06:00
Philip O'Toole 25f957c5c6 Only call Stop on non-nil batchers 2015-10-14 08:55:06 -07:00
Philip O'Toole a938cd3dee openTSDB Open should complete before Close runs 2015-10-14 08:55:06 -07:00
Philip O'Toole 3907656cc2 Add README for UDP service
Fixes issue #4041.
2015-10-14 08:30:10 -07:00
Philip O'Toole f298e88b39 Auto-create UDP service database
All other services operate like this, so make UDP service consistent.
2015-10-14 08:30:09 -07:00
Sean Beckett 5ab86f7578 Update README.md 2015-10-13 16:56:37 -07:00
Sean Beckett ed7b9f7485 tag name --> tag key 2015-10-13 16:41:57 -07:00
Daniel Morsing 822af73f88 implement continuous queries as regular execs of into queries.
Now that we have into queries, we can implement them as regular
queries that are just run on a timer.
2015-10-13 15:51:19 +00:00
Sergey Kamardin d25e264009 Update handler.go
Add `Access-Control-Expose-Headers` for `ping` endpoint clients be able to retrieve `X-Influxdb-Version` and `Date` from the server.
2015-10-12 18:05:18 +03:00
Philip O'Toole 37cf9a1610 Deletion while iterating is OK in Go 2015-10-09 16:30:20 -07:00
Philip O'Toole f12470a99e If there are no HH segments, then nothing to purge 2015-10-09 14:29:21 -07:00
Philip O'Toole c06ac8f94c Don't add a new segment every purge check
Everytime the purge check was running, a new segment was being added.
This meant the list of almost-empty files in the HH directories would
grow continually.
2015-10-09 14:26:47 -07:00
Philip O'Toole 657aa5a134 Add README for collectd 2015-10-09 09:15:22 -07:00
Philip O'Toole b009f25e3d Delete queues for inactive nodes
Deletion only takes place if all data in the queue is older than the
configured time.
2015-10-08 20:34:24 -07:00
Philip O'Toole 5b0a8ed306 HH should not process dropped nodes 2015-10-08 18:23:12 -07:00
Cameron Sparr 2add55107e Fix graphite parser merge error, nargs 2015-10-08 11:15:03 -06:00
dgnorton a9bf213076 Merge pull request #3484 from dawbs/dawbs-fix-3429
Bugfix for #3429 String representations of RegexLiterals generated in…
2015-10-08 13:12:10 -04:00
Cameron Sparr 3bea25b428 graphite parser: apply tags from the Parser on the template 2015-10-08 10:56:13 -06:00
Nick Dawbarn 136dbef0e7 Formatting fixes 2015-10-08 19:41:36 +10:00
Nick Dawbarn 26f6d00668 Bugfix for #3429 String representations of RegexLiterals generated in influxql/ast.go add the / char as a start and end delimiter, but does not escape any / characters that may exist with the regex 2015-10-08 19:41:36 +10:00
Cameron Sparr 73a630dfa6 graphite parser: apply tags from the Parser on the template 2015-10-07 23:19:29 -06:00
Rob Wilson f3e3bf7a0e typo 2015-10-07 21:25:49 +01:00
Rob Wilson 5815e0b0ee updated documentation 2015-10-07 21:24:05 +01:00
Rob Wilson d8ac746703 correct formatting 2015-10-07 20:35:05 +01:00
Rob Wilson 5fd8777c56 add tests 2015-10-07 20:32:10 +01:00
Rob Wilson a27186fb7a raise exception when field keyword is specified multiple times 2015-10-07 20:31:46 +01:00
Rob Wilson bcd6c06173 Merge remote-tracking branch 'upstream/master' into graphite-template-custom-field
Conflicts:
	services/graphite/parser.go
2015-10-07 17:48:34 +01:00
Philip O'Toole 44d52ac138 Fully lock HH node queue creation
I believe this change address the issues with hinted-handoff not fully replicating all data to nodes that come back online after an outage.. A detailed explanation follows.

During testing of of hinted-handoff (HH) under various scenarios, HH stats showed that the HH Processor was occasionally encountering errors while unmarshalling hinted data. This error was not handled completely correctly, and in clusters with more than 3 nodes, this could cause the HH service to stall until the node was restarted. This was the high-level reason why HH data was not being replicated.

Furthermore by watching, at the byte-level, the hinted-handoff data it could be seen that HH segment block lengths were getting randomly set to 0, but the block data itself was fine (Block data contains hinted writes). This was the root cause of the unmarshalling errors outlined above. This, in turn, was tracked down to the HH system opening each segment file multiple times concurrently, which was not file-level thread-safe, so these mutiple open calls were corrupting the file.

Finally, the reason a segment file was being opened multiple times in parallel was because WriteShard on the HH Processor was checking for node queues in an unsafe manner. Since WriteShard can be called concurrently this was adding queues for the same node more than once, and each queue-addition results in opening segment files.

This change fixes the locking in WriteShard such the check for an existing HH queue for a given node is performed in a synchronized manner.
2015-10-07 02:33:43 -07:00
Philip O'Toole 5b0767c30b EOF is OK in HH processor 2015-10-07 01:56:55 -07:00
Philip O'Toole 8b49c37120 Count HH errors 2015-10-06 20:49:40 -07:00
Philip O'Toole 5d5515a497 If HH can't unmarshal a block, skip that block 2015-10-06 20:49:40 -07:00
Cameron Sparr 883d32cfd0 Add public function to graphite parser to apply template 2015-10-06 17:42:36 -06:00
Paul Dix bb398daf75 Updates based on @otoolp's PR comments 2015-10-05 20:09:56 -04:00
Jason Wilder 5d9b89d601 Disable copier test
Not implemented for tsm1 engine
2015-10-05 20:09:56 -04:00
Paul Dix 7555ccbd70 WIP: engine work 2015-10-05 20:06:21 -04:00
Philip O'Toole 2ac0357406 Support dropping non-Raft nodes 2015-10-04 00:19:52 -07:00
Philip O'Toole d74e0690c7 Revert "Merge pull request #4233 from influxdb/drop-server"
This reverts commit 0bdb36f6dc, reversing
changes made to 3085fbc138.
2015-10-02 08:39:57 -07:00
Cory LaNou f50813460e protobuf update.. :-( 2015-10-01 15:39:15 -05:00
Philip O'Toole 8a1e5a9e53 Clamp initial value of HH retry interval
This could happen due to misconfiguration, so do something sensible in
that case.
2015-10-01 12:04:33 -07:00
Philip O'Toole 878f776403 Exponential backoff if any hinted-handoff fails 2015-09-30 21:27:13 -07:00
Philip O'Toole 4eba2c1725 Add config support for max HH retry interval 2015-09-30 21:10:03 -07:00
Philip O'Toole 235714755c HH processor-level stats
This change maintains stats on a per-shard and per-node basis.
2015-09-28 18:39:39 -07:00
Philip O'Toole 14db3ce9f5 Add service-level stats for hinted-handoff 2015-09-28 18:08:35 -07:00
Philip O'Toole a196d3663a Allow configuration of UDP retention policy
Fixes issue #4529
2015-09-28 15:17:56 -07:00
Philip O'Toole 49a70d0fca Merge pull request #4238 from influxdb/hh_control
Fully disable hinted-handoff service if requested
2015-09-28 12:11:18 -07:00
Philip O'Toole a4a8fa0ff0 Fully disable hinted-handoff service if requested
Without this change if hinted-handoff was disabled the service would
correctly reject writes, but it would process any data sitting in
hinted-handoff queues. With this change the service is completely
disabled.
2015-09-25 18:03:43 -07:00
Philip O'Toole 9de3125f6b Graphite TCP should not block system shutdown
With this change Graphite TCP connections are tracked on a per-service
basis. This allows a closing Graphite service to first shutdown any
active connections, thereby unblocking the rest of shutdowm.

This work exposed small shortcomings with the existing Diagnostics
system and that code has alse been tweaked.

Fixes issue #4017
2015-09-24 14:08:38 -07:00
Antonio Murdaca 49c0b6ea73 Fix go vet warnings
This patch fixes the following go vet warnings:

```
services/continuous_querier/service.go:326: influxql.Statements
composite literal uses unkeyed fields
exit status 1
services/httpd/handler_test.go:145: models.Rows composite literal uses
unkeyed fields
services/httpd/handler_test.go:146: models.Rows composite literal uses
unkeyed fields
services/httpd/handler_test.go:165: models.Rows composite literal uses
unkeyed fields
services/httpd/handler_test.go:166: models.Rows composite literal uses
unkeyed fields
services/httpd/handler_test.go:187: models.Rows composite literal uses
unkeyed fields
services/httpd/handler_test.go:188: models.Rows composite literal uses
unkeyed fields
exit status 1
```

Signed-off-by: Antonio Murdaca <runcom@linux.com>
2015-09-21 15:28:54 +02:00
Rob Wilson ef35d6dcc2 formatting 2015-09-21 12:26:43 +01:00
Rob Wilson 27c1cc23fd Working prototype.. 2015-09-21 12:18:19 +01:00
Rob Wilson 9121b422f8 comment out tests for now.. 2015-09-21 10:47:42 +01:00
Rob Wilson 20e4fdfa9a allow specifying fieldname in graphite template 2015-09-20 21:17:50 +01:00
Cory LaNou 72f6f7d268 Merge pull request #4134 from influxdb/issue-3447
Refactor Points and Rows to dedicated packages
2015-09-17 15:27:48 -05:00
Cory LaNou 38cb7b49de Mising defer in httpd recovery. fixes #4124 2015-09-17 09:37:27 -05:00
Cory LaNou ba830be3b9 actually move influxql.Row* -> models.Row* 2015-09-16 16:32:50 -05:00
Cory LaNou d19a510ad2 refactor Points and Rows to dedicated packages 2015-09-16 15:33:08 -05:00
Philip O'Toole d538829b4c Enhance openTSDB logging and stats 2015-09-09 13:30:11 -07:00
Philip O'Toole fef20c77b2 Cleanly terminate openTSDB connection on EOF
This is not really an error, so don't log it.
2015-09-09 13:01:13 -07:00
Philip O'Toole 02fcaf853d Add node re Graphite configuration
[ci skip]
2015-09-08 23:22:34 -07:00
Philip O'Toole 519a30a463 Add note on openTSDB batching
[ci skip]
2015-09-08 23:19:17 -07:00
Philip O'Toole 24aca5611a Add batch-pending control to openTSDB input 2015-09-08 19:35:42 -07:00
Philip O'Toole 95530e1623 Set UDP input defaults if not set 2015-09-08 19:32:20 -07:00
Philip O'Toole 5373f263a3 Add pending control to batcher
With this change, the generic batcher used by many inputs can now be
buffered. Testing shows that this performance of the Graphite input by
10-100%, with the biggest improvements at lower numbers of connections.
2015-09-08 19:32:00 -07:00
Philip O'Toole e38a204afc Merge pull request #4043 from influxdb/opentsdb_batching
Add batching and stats to openTSDB input
2015-09-08 19:27:35 -07:00
Philip O'Toole 1ce5187b66 Merge pull request #4049 from influxdb/udp_stats
Add stats to the UDP input
2015-09-08 19:18:17 -07:00
Philip O'Toole 9677a0faab Add collectd stats 2015-09-08 19:07:47 -07:00
Philip O'Toole 27932409b0 Add stats to the UDP input 2015-09-08 18:48:35 -07:00
Philip O'Toole 817328d378 Add basic stats to the CQ service 2015-09-08 18:17:20 -07:00
Philip O'Toole 349ba8b307 Add batching and stats to openTSDB input 2015-09-08 16:19:50 -07:00
Jason Wilder 73510a0a68 Fix invalid time stamp in graphite metric causes panic
If a timestamp was larger than the max epoch value was sent via
graphite it would cause the timestamp to overflow when it was
marshaled/unmarshaled back from the raft log.  The overflow cause
the shard group to get created with the wrong timestamp which cause
a panic when writing the point.  The panic was caused because the
timestamp that were supposed to exists in a map created by MapShards
did not actually exist so a nil ShardGroup was used.

The change prevents creating the point with an invalid timestamp.  Since
graphite using a timestamp in seconds, the maximum range is known and
can be prevented.  This also adds a check for the minimum range as well.

Fixes #3785
2015-09-08 10:07:47 -06:00
Philip O'Toole 332ce6481d Removed unused Graphite NewConfig
This function is not helpful for sections of the config that support
multiple instances.
2015-09-08 08:32:19 -07:00
Philip O'Toole bbc103305b Support multiple Graphite inputs
Fixes issue #3636
2015-09-06 21:33:46 -07:00
Philip O'Toole fa29e12222 Shutdown UDP Graphite on SIGTERM
Service.Close() had no way of closing the UDP Conn. This change makes
the UDP an attribute of the server, so Close() can access it.
2015-09-05 00:30:59 -07:00
Philip O'Toole 579e2a250c Add stats to httpd package 2015-09-04 12:37:59 -07:00
Philip O'Toole 3df898bd90 Merge pull request #3987 from influxdb/global_expvar_hookup_diagnostics
Use expvar statistics directly
2015-09-04 11:13:17 -07:00
Philip O'Toole 89bc392ec4 Access expvar directly from monitor
expvar map is already global so access it directly. This simplifies the
code and makes it much eaisier to use from other modules.
2015-09-04 09:45:24 -07:00
Philip O'Toole cf5a655249 Don't precreate shard groups entirely in past
Fixes issue #3722
2015-09-04 08:31:50 -07:00
Philip O'Toole 6ad35e23e9 Integrate code review feedback 2015-09-03 20:50:54 -07:00
Philip O'Toole d58532d844 Add Graphite diagnostics
Graphite diagnostics currently show TCP connections.
2015-09-03 20:50:54 -07:00
Philip O'Toole e07432c59f Implement diagnostics support
This change adds support for diagnostics by decomposing the existing
interface into two interfaces -- one for stats, and the other for
diags. It also adds some basic monitor of system, network, and the Go
runtime.
2015-09-03 20:50:54 -07:00
David Norton dce666e757 fix #3979: fix race in CQ service 2015-09-03 19:55:40 -04:00
Ben Johnson deff06f850 add copier service
This commit adds the copier service which allows one server to
copy shards from another server. This will be used for moving
shards in the cluster.
2015-09-03 13:07:35 -06:00
David Norton 0cb9618d6d fix CQ intoDB() 2015-09-03 09:07:57 -04:00
David Norton d466b19388 update CQ service unit tests 2015-09-03 07:12:15 -04:00
David Norton 66001cfbb5 fix #2555: add integration tests for CQs 2015-09-03 07:12:15 -04:00
David Norton 021a6f5453 rename CQ tests 2015-09-03 07:12:15 -04:00
David Norton 99a22c174b fix #2555: add backreference in CQs
Add new query syntax to allow the following in CQs:

INTO "1hPolicy".:MEASUREMENT
2015-09-03 07:12:15 -04:00
Philip O'Toole 4e2ee1ea70 Rename MonitorService to just Monitor
monitor is not a service, it has more in common with meta, since it
provides functionality to the query layer. This names makes this
clearer.
2015-09-02 15:07:30 -07:00
Philip O'Toole 366c0115f9 Serve expvar information from HTTP package 2015-09-01 15:22:37 -07:00
Philip O'Toole 9df17409d3 Use monitor service with Graphite 2015-09-01 15:21:36 -07:00
Philip O'Toole d87e668c78 Remove obsolete monitoring code 2015-09-01 15:03:52 -07:00
Philip O'Toole d771612718 Set default retention check interval to 30 minutes
Since the minimum retention period is 1 hour, checking every 10 minutes
seems excessive and generates noise in the logs.
2015-08-27 16:08:03 -07:00