This will be changed in a second

Now it is possible to make the second argument (query parameters)
optional. I.e. now we support 'pico.sql([[select * from t]])' as
well as 'pico.sql([[select * from t where a = ?]], {1})'.

Egor Ivkov authored 1 year ago and Егор Ивков committed 1 year ago

They should cover all the cases specified in the CaS Design Document

Current commit introduces distributed SQL in picodata (we use
'sbroad' library). Now it is possible to execute cluster-wide
SQL with 'pico.sql([[query]], {params})' lua method.

Current limitations:
1. DDL is still not implemented in picodata, so space sharding
   is mocked by '_pico_space' space (we'll remove it later).
2. We still have duplicating symbols of the Tarantool module in
   sbroad and in picodata rlibs.
3. We need to reuse piocdata's sharding functions and handle the
   case when different spaces in the query have different sharding
   functions.

Co-authored-by: Georgy Moshkin <gmoshkin@picodata.io>

Dmitry Travyan authored 1 year ago and Yaroslav Dynnikov committed 1 year ago

This patch introduces `picodata connect` CLI command. It allows
connecting to interactive lua console of a picodata instance via iproto
The whole implementation comes down to invoking a lua script with
`require("console").start()`.

The main focus is on the snapshot recovery scenarios (bootstrap /
cathcup) interleaving with explicit CaS operations, rather than on
leadership changes and joins.

Consider the case when a client requests `index: X, term: T`, but
`term(X) != T`. Picodata used to check all entries in the range `[X+1,
last]`, but it didn't check the index `X` itself.

This patch provides this one extra check. Thereby it makes sure that
all entries in the range `[X, last]` belong to the same term `T ==
current_term`.

Refactor corresponding tests accordingly

Georgy Moshkin authored 2 years ago and Yaroslav Dynnikov committed 1 year ago

This patch includes:

- Tarantool stored procedure `proc_cas`.
- Lua API `pico.cas`.
- Integration tests.

Now the function returns `first_index` instead of the number of entries
deleted - it has no practical purpose.

This patch adds a missing test for `pico.raft_compact_log` Lua API.
Also, it changes the API to be more alike other relevant Lua functions:
`raft_propose_*`, `raft_read_index`, `raft_timeout_now`.

Yaroslav Dynnikov authored 1 year ago and Yaroslav Dynnikov committed 1 year ago

Adopted styleguide is "actual == expected", not vice versa.

Bumping `tarantool-module` fixes the `fiber::csw()` performance issue
that caused flaky tests in CI.

Also this patch increases the `test_sigsegv_child` timeout. The test is
flaky too, let's see if it helps.

It could happen that i1 persists an extra entry before i2 gets it, and
if i2 starts requesting a vote over and over, it won't get that entry.

NOTE: there's still a bug, because we set each new replicaset's weight
to 1 (even ones which don't satisfy replication factor)
before the bucket distribution is bootstrapped. But there must be at
least one replicaset with non zero weight in order for vshard.*.cfg to
work.

A potential solution would be to only configure vshard once a replicaset
is filled up.