Add new config::AlterSystemParameters struct which is responsible
for definitions of system parameters, their default values and mapping
to sbroad type.

This is a big step towards moving all system parameters from
_pico_property into another system table.

Note that after this merge request, we no longer validate whether the format
of the created object matches with the schema. Consequently, we do not
suppress errors if formats match anymore.

Before:
```sql
CREATE USER u WITH PASSWORD 'Passw0rd'
1
CREATE USER u WITH PASSWORD 'Passw0rd'
0
CREATE USER u WITH PASSWORD 'DifferentPassw0rd'
error: user already exists with different auth method
```

Now:
```sql
CREATE USER u WITH PASSWORD 'Passw0rd'
1
CREATE USER u WITH PASSWORD 'Passw0rd'
error: user u already exists
CREATE USER u WITH PASSWORD 'DifferentPassw0rd'
error: user u already exists
CREATE USER IF NOT EXISTS u WITH PASSWORD 'DifferentPassw0rd'
0
```

godzie44 authored 6 months ago and Georgy Moshkin committed 6 months ago

Closes #664

This commit restores the user name in the ALTER USER command to "Dave" after
it was mistakenly changed to "DAVE", a nonexistent user. This change
broke the test logic, causing the test to erroneously verify granting
privileges to a nonexistent user, rather than confirming that we can
grant default privileges. The previous implementation resulted in noop,
so this went unnoticed until we started validating user existence.

Here is the commit changing Dave to DAVE:
https://git.picodata.io/picodata/picodata/picodata/-/commit/ecbb520b559ceb7f2dbe7ae0462c8d062faf1589

This reverts commit d7e93cfc.

CAS predicates should always contain the current raft term, because this
is what's explicitly checked in proc_cas.

Note that this will sometimes result in a EntryTermMismatch error
because the latest applied entry may have a different term in case the
election has started but not finished yet.

We could add this check to all the client cas precondition checks, but
it's not a big deal, because we handle the situation correctly anyways.

Rolled back as in raft log roll back, which can happen when a leader
changes to one who doesn't have the entry we persisted.
Not to be confused with DdlAbort.

feat: support ALTER SYSTEM. It allows to change whitelisted properties from _pico_property table. Currently, changes affect all tiers. Specifying tier will result in an error.

Co-Authored-by: Dmitry Rodionov <d.rodionov@picodata.io>

Co-Authored-by: Emir Vildanov <e.vildanov@picodata.io>

Arseniy Volynets authored 7 months ago and Arseniy Volynets committed 6 months ago

- update sbroad submodule to commit with
new normalization rules
- update integration tests

Dmitry Rodionov authored 7 months ago and Yaroslav Dynnikov committed 7 months ago

Previously function was ~700 loc. The patch splits out the logic of
conversion from ir node to operation. Previously there were two stages,
initial validation that constructed intermediate structure and then
intermediate structure was transformed into Op potentially several times
in case of retries. The patch removes intermediate conversion by
direct mapping from ir node to op. It makes reading more convenient
since there is no need to jump back and forth between initial checks and
actual Op construction. Another benefit is that some checks need to be
repeated on each retry since retries originate from conflicts and
conflicts change system state. So in newer state operation may no longer
make sense. By coupling both operations together there is no way a check
can be missed in the retry phase.

We still do some cloning here and there. We can probably improve on that
by constructing Op once and then adjusting it in place for conflicts but
since retries should be rare in practice it shouldnt be a problem.

Additionally we do a lot of conversion between SmolStr and String. We
need to use SmolStr in more placese inside picodata so we dont need to
do that.

- when doing global dml sql query, there is error
like user don't have access to 'raft_state` space.
Fix this by doing parts of pipeline that access
system spaces under admin

- update sbroad submodule to commit with helper
patches for dml on global tables
- add support for sql dml insert/update/delete
on global tables. Insert on conflict is not
supported yet
- Current implementation uses a single batch of
dml commands via CAS. No additional predicates
are used in CAS, so no isolation level is guaranteed
even when query only refers to global tables