Feature/dynamo store #221
Feature/dynamo store #221
Conversation
NIGHTLY checkin
Conflicts: src/github.com/mozilla-services/pushgo/simplepush/aws.go
Ok, it doesn't cache a LOT, but it might reduce some time.
…/dynamo_store Conflicts: Makefile src/github.com/mozilla-services/pushgo/simplepush/mock_proprietary_ping_test.go src/github.com/mozilla-services/pushgo/simplepush/mock_socket_test.go src/github.com/mozilla-services/pushgo/simplepush/mock_storage_test.go src/github.com/mozilla-services/pushgo/simplepush/mock_worker_test.go
(needs test)
* DRY the tests * Convert some strings to CONST labels * Added DropAll and test (for APNs work) * changed "created" tag to "modified" ** modified modification modifiers to modify "modification" on modification.
| done := make(chan bool) | ||
| timeout := time.After(s.statusTimeout) | ||
| go func() { | ||
| for { |
There was a problem hiding this comment.
What does the select statement and <-done do?
There was a problem hiding this comment.
It performs a loop (with the time.Sleep below). I'll grant that it's a little confusing with the timeout "done" below (which is triggered on a close).
This function was copied from the dynamodb_test lib.
https://github.com/goamz/goamz/blob/master/dynamodb/dynamodb_test.go#L102
There was a problem hiding this comment.
Oh, I see...done is unbuffered, so the send on line 105 will block until it's matched by a receive on line 113. Otherwise, if the receive times out, done will be closed, signaling the loop to exit on the next turn. That took a few tries to parse correctly. 😁
| @@ -19,7 +19,8 @@ github.com/ianoshen/gomc 7b9f299f292d3dd707fe2749d966968c9bf1e128 | |||
| github.com/kitcambridge/envconf fdc1968532255bdb56182329cdaa1e5b9aa6a6ac | |||
| github.com/smartystreets/goconvey 8298bc7d36389ffd3e57b85d9797850d8c2382a9 | |||
| github.com/jacobsa/oglematchers 4fc24f97b5b74022c2a3f4ca7eed57ca29083d3e | |||
|
|
|||
| github.com/jrconlin/goamz/aws 7d8ddc499fd11cfc58fc2f6a4dcc7e0ff6f930ca | |||
| github.com/jrconlin/goamz/dynamodb 7d8ddc499fd11cfc58fc2f6a4dcc7e0ff6f930ca | |||
There was a problem hiding this comment.
goamz has applied your patches, we should switch back to their repo.
| WriteProv int64 `toml:"write_cap_units" env:"write_cap_units"` | ||
|
|
||
| // Region that the dbstore is in | ||
| Region string `toml:"region" env:"region"` |
There was a problem hiding this comment.
toml/env shouldn't be needed since the name is identical.
| Access: "MISSING_ACCESS_KEY", | ||
| Secret: "MISSING_SECRET_KEY", | ||
| ReadProv: 1, | ||
| WriteProv: 1, |
There was a problem hiding this comment.
I thought minimum write/read units were 10/50 respectively. Will AWS just up it to that if you say 1?
There was a problem hiding this comment.
Well, 1/1 is a lot less than the free tier of 25/25. I was using that as a config mostly because I was just doing simple function testing. I'll bump these values to the free provisioning layer.
There was a problem hiding this comment.
Ah! No worries, it can be left there. I had thought the min. was 10.
|
r- pending some nits, doc changes, panic condition, dynamodb operation changes. |
|
One other note, every read command by default is eventually consistent, we need strongly consistent, which is an additional flag to add to every read/query request. |
| // value (it only updates if the previous value matches) | ||
| // this is a problem for new fields or running statelessly. | ||
| // It does mean that there's a chance of a delayed packet decrementing | ||
| // a version or otherwise causing an issue. |
Add handler for aws DynamoDB storage.
This is using a custom patch for the goamz pending landing goamz goamz/goamz#67
The tests for this package run through a home-grown mock. This mock stubs out the AWS calls in favor of a in-memory checker. This prevents the need to run a secondary server (goamz).
The library will also auto-create the dynamodb tables if not present. The schema includes a global secondary index based off the
This patch is for 1.6+