distributed-database

"found" by the @discordapp troops the hard way: https://status.discordapp.com/incidents/62gt9cgjwdgf
pinging @zorkian who pointed it out to me.

trivial to repro, this is against a 3.3.13 on fedora-31.

$ etcdctl set /baa schnorp
$ etcdctl get /baa
schnorp
$ echo -ne 'PUT /v2/keys/baa HTTP/1.1\r\nHost: boo\r\nContent-Length: 123\r\n\r\n' | nc localhost 2379
HTTP/1.1 200 OK
...
$ et

Is your feature request related to a problem? Please describe.
When restoring a backup, a user may run into errors due to specific tables/descriptors. In some cases, the user may not care about that table that couldn't restore (e.g. system.jobs or a temp table), however, that faulty table/descriptor ends up blocking the whole restore.

Describe the solution you'd like
Allow the user to

The first constructor of SelectStatementContext are used for test case only.

Please refactor the test cases to use the second constructor and remove the first one.

fdbcli flags currently contain a mixture of _ and - (probably for historic reasons). I would propose to unify the flags and only use either _ or - personally I tend to hyphens like recommended in the POSXI standard (https://www.gnu.org/software/libc/manual/html_node/Argument-Syntax.html). Current fdbcli help output:

FoundationDB CLI 6.2 (v6.2.29)
usage: /usr/bin/fdb/6.2/fdbcli

Hello Philip!
I think there is an issue with this part of the code of rqlite (store/store.go).

func (s *Store) Database(leader bool) ([]byte, error) {
	if leader && s.raft.State() != raft.Leader {
		return nil, ErrNotLeader
	}
	// Ensure only one snapshot can take place at once, and block all queries.
	s.mu.Lock()
	defer s.mu.Unlock()

	f, err := ioutil.TempFile("", "rqlilte-snap-

Flush/Compact requests to the tServer are rpcs that are expected to take a long time. It does not make sense for these to use the same deadline (20s) and num_retries as other async_rpc_tasks.

We should have these tasks override these settings; similar to what BackfillChunk does to give itself more time; and also to disble/reduce retries in the case of flush/compact.

Without having this, we

In the past a conversion through Number "interface" has proven to be able to mask bugs in the code.
We should remove these conversions from io.trino.spi.type.TypeUtils#writeNativeValue.

Use case:

1.) A user may want to backup all tables but no metadata like users, privileges, etc. without explicitly defining each table inside the CREATE SNAPSHOT statement.

2.) A user may want to transfer users & privileges, custom analyzers or user-defined-functions from one cluster to another without backing up a complete cluster including all data (tables).

*Feature description

Currently we don't have any mechanism to limit the maximum number of clients that could be handled simultaneously.

This feature should be designed properly. Here is some clue: https://redis.io/topics/clients#maximum-number-of-clients

Now that we support more than 1 data type, we should include the values data type in cache nodes. This give us a way of easily identifying what type of data a key points to e.g. string, queue (and more types in the future)

Currently a cache node object consists of a Key, a Value and a TTL. We should also add a field specifying the type of the Value