…e.proto.

Implement ParallelRunner and Checkpointer. Create a simple test for
ParallelRunner.

workflow. Complete the unit test and E2E test for happy path.

unit test to verify this function. Implemented Horizontal Resharding
workflow and tested in unit test and e2e test for the happy path.

It's just exposing the library methods into a binary, and using it in
python tests.

And a unit test that validates it.

With tests. Also, the server side needs to not use a buffer read for the
first handshake packet, so it doesn't buffer TLS negociation packets as
well.

for change on node.go in workflow folder.

This happened only when shards were included as well.

Upon initial rebuild for a new keyspace, also call RebuildVSchema

Created the Code enum to match grpc error codes, and renamed
all previous error codes to not have a name clash. Also renamed
The older field to LegacyCode.

vtrpcpb.ErrorCode is used only at the grpc boundary for
legacy purposes.

* Renamed old error names to legacy.
* Changed all clients to handle legacy and new codes.
* Added tests for the conversions.

Tests were using fatal/retry style error handling.
Those are now updated to use vtrpcpb.Code.

Also fixed error code priorities in the error aggregation function.

BUG=32851872

All VTGate error types have been consolidated to use
vterrors.
Also, /debug/vars error reporting has been consolidated.
Theres is now only one error var, but it has a fourth
category which is the Code. We can consider dropping
one of the other categories as the number of combinations
may be too high.

The wrapper functionality now takes over the retry decision
because it's better suited to make it. For example, a streaming
query should not be retried if it's already sent results.

Also delete TabletError

vtgate/buffer: Fix bug that keyspaces were not logged at startup.

horizontal resharding workflow. First round comments resolved.
(addressing race condition warnings)

workflow: resharding workflow: Implement checkpointing.

This is a follow-up fix for vitessio#2495.

We were not able to fix Yipei's workstation such that it would use the
same protobuf generator as Travis and our setups does. Therefore, I'm
re-generating the files separately on my machine.

proto: Regenerate protobuf files.

I've also regenerated all protobuf files.

Note that the new version has added getters for primitive types. Do NOT
use these getters and instead always use the field names instead.

Cipher must be a legal UTF8 string.

vendor: Pin the Protobuf Go generator to the latest version.

In the process, allow a vttest user to specify extra my.cnf.

just skip any unparseable rbr events

…_throttler

Moved creation of topo.Server outside of tx_throttler.

retry logic.

…ecked.

workflow: Remove RetryController object and simplify ParallelRunner

…ys_nil

tabletserver/txthrottler: Fix bug that config update error was not ch…

SSL: mysqlconn TLS support, better tests.

And a config-file based implementation.
Using it in all tests.

workflow: Modify Factory.Init to pass Manager object.

Old files could not be watched if the server had compacted its data.

Use the Get version as a start, not the object version.

Etcd{,2} fix for watches.

Addressed review comments. I've changed the error codes for
cases we've agreed on. I've left the more contentious ones
unchanged for now.

BUG=32851872

And connecting it to AuthServer.
Unit tests cover almost all cases (but not all pre-MySQL 4.1 cases).

Pass MoreExecutors.directExecutor to Futures.transformAsync.

Seems to happen on Travis.

…sts.

Canceled requests, e.g. those with a deadline shorter than the failover duration, would not return their buffer pool slot. Eventually, all slots would have leaked and the buffer would start and stop buffering as usual but reject all requests immediately because it assumed there are other pending failovers which are holding the needed slots.

Previously, the code path for canceled requests differed from the common method unblockAndWait() which would unblock a request and also release its buffer pool slot after the request finished its retry. This made this oversight possible. I've changed this now: canceled requests also use unblockAndWait() now.

The code was also out of sync with the documentation for WaitForFailoverEnd(): The RetryDoneFunc() must not be returned when the function returns an error as well. However, for canceled requests both were returned.

vtgate/buffer: Fix leakage of buffer pool slots due to canceled requests.

* Disallow creation of error with OK as status code.
  Auto-convert to INTERNAL and log error.
* Handle context errors as special-case.
  Return more appropriate error code instead of UNKNOWN.
* Also log UNAVAILABLE errors in vtgate because they're
  often due to unreachable vttablets.

…eturned by MySQL

Also adding an end-to-end test to validate the connector and the caller
id.

…, add comments and clean up some helper methods

…length

…sed_throttler_doc

Added documentation for the replication-lag-based-throttler

The root cause of the deadlock is that message manager calls
into tabletserver, which calls back into it.
This can cause deadlocks and Close can hang forever.

BUG=35763775

Auth work for MySQL Server code.

modify the unit tests of resharding workflow.

NOTE: This is an automated export. Changes were already LGTM'd internally.

…l_err

Do not return OK error code for buffer errors.

suggestion.

Add flags to kubernetes guestbook app.

…nager

workflow: Modify the unit test for resharding workflow. Control the workflow through manager.

If we don't do this, monitoring frameworks may not correctly calculate rates for the first failover of the shard because they see a transition from "no value for this label set (NaN)" to "a value".

* Sandbox reliability updates and cleanup.

* Address comments.

* Address more comments.

vtgate/buffer: Always initialize all stat variables.

* Updating the example with New Vschema format

* Making VSchema changes in sharding-kubernetes.html

Vschema doc fix

Makefile: fix docker build mysql56 bug

It has not been working for a while and I was not able to fix it.

Additionally, it recently started to flake e.g. we saw this error:

> Bad response status from coveralls: 422 - {"message":"Couldn't find a
repository matching this job.","error":true}

https://travis-ci.org/youtube/vitess/jobs/206464990

Let's remove it and replace it with something else at a later point.

Adds decoding of strings and string-like blobs based on the charset returned by MySQL

Since go/vt-pull/2608 it is always initialized and broke this test.

travis: Stop publishing coverage to coveralls.io.

…x_export

test: vtgate_buffer.py: Ttest for the varz value and not just existence.

Removed the old unused query_splitter.go and related test

update success flag

This is recommended by the Guava docs:

"For field types and method return types, you should generally use the immutable type (such as ImmutableList) instead of the general collection interface type (such as List). This communicates to your callers all of the semantic guarantees listed above, which is almost always very useful information."

From: https://google.github.io/guava/releases/21.0/api/docs/com/google/common/collect/ImmutableCollection.html