Move replica-oriented classes to the replica namespace. The main
classes moved are ::database, ::keyspace, and ::table, but a few
ancillary classes are also moved. There are certainly classes that
should be moved but aren't (like distributed_loader) but we have
to start somewhere.
References are adjusted treewide. In many cases, it is obvious that
a call site should not access the replica (but the data_dictionary
instead), but that is left for separate work.
scylla-gdb.py is adjusted to look for both the new and old names.
The database, keyspace, and table classes represent the replica-only
part of the objects after which they are named. Reading from a table
doesn't give you the full data, just the replica's view, and it is not
consistent since reconciliation is applied on the coordinator.
As a first step in acknowledging this, move the related files to
a replica/ subdirectory.
The header file <seastar/net/ip.hh> is a large collection of unrelated stuff, and according to ClangBuildAnalyzer, takes 2 seconds to compile for every source file that included it - and unfortunately virtually all Scylla source files included it - through either "types.hh" or "gms/inet_address.hh". That's 2*300 CPU seconds wasted.
In this two-patch series we completely eliminate the inclusion of <seastar/net/ip.hh> from Scylla. We still need the ipv4_address, ipv6_address types (e.g., gms/inet_address.hh uses it to hold a node's IP address) so those were split (in a Seastar patch that is already in) from ip.hh into separate small header files that we can include.
This patch reduces the entire build time (of build/dev/scylla) by 4% - reducing almost 10 sCPU minutes (!) from the build.
Closes#9875
* github.com:scylladb/scylla:
build performance: do not include <seastar/net/ip.hh>
build performance: speed up inclusion of <gm/inet_address.hh>
In a previous patch, we noticed that the header file <gm/inet_address.hh>,
which is included, directly or indirectly, by most source files,
includes <seastar/net/ip.hh> which is very slow to compile, and
replaced it by the much faster-to-include <seastar/net/ipv[46]_address.hh>.
However, we also included <seastar/net/ip.hh> in types.hh - and that
too is included by almost every file, so the actual saving from the
above patch was minimal. So in this patch we replace this include too.
After this patch Scylla does not include <seastar/net/ip.hh> at all.
According to ClangBuildAnalyzer, this reduces the average time to include
types.hh (multiply this by 312 times!) from 4 seconds to 1.8 seconds,
and reduces total build time (dev mode) by about 3%.
Some of the source files were now missing some include directives, that
were previously included in ip.hh - so we need to add those explicitly.
Signed-off-by: Nadav Har'El <nyh@scylladb.com>
"
Like flat_mutation_reader_from_fragments, this reader is also heavily
used by tests to compose a specific workload for readers above it. So
instead of converting it, we add a v2 variant and leave the v1 variant
in place.
The v2 variant was written from scratch to have built-in support for
reading in reverse. It is built-on `mutation::consume()` to avoid
duplicating the logic of consuming the contents of the mutation. To
avoid stalls, `mutation::consume()` gets support for pausing and
resuming consuming a mutation.
Tests: unit(dev)
"
* 'flat_mutation_reader_from_mutations_v2/v2' of https://github.com/denesb/scylla:
flat_mutation_reader: convert make_flat_mutation_reader_from_mutation() v2
flat_mutation_reader: extract mutation slicing into a function
mutation: consume(): make it pausable/resumable
mutation: consume(): restructure clustering iterator initialization
test/boost/mutation_test: add rebuild test for mutation::consume()
Since this reader is also heavily used by tests to compose a specific
workload for readers above it, we just add a v2 variant, instead of
changing the existing v1 one.
The v2 variant was written from scratch to have built-in support for
reading in reverse. It is built-on `mutation::consume()` to avoid
duplicating the logic of consuming the contents of the mutation.
A v2 native unit test is also added.
The gc_grace_seconds is a very fragile and broken design inherited from
Cassandra. Deleted data can be resurrected if cluster wide repair is not
performed within gc_grace_seconds. This design pushes the job of making
the database consistency to the user. In practice, it is very hard to
guarantee repair is performed within gc_grace_seconds all the time. For
example, repair workload has the lowest priority in the system which can
be slowed down by the higher priority workload, so that there is no
guarantee when a repair can finish. A gc_grace_seconds value that is
used to work might not work after data volume grows in a cluster. Users
might want to avoid running repair during a specific period where
latency is the top priority for their business.
To solve this problem, an automatic mechanism to protect data
resurrection is proposed and implemented. The main idea is to remove the
tombstone only after the range that covers the tombstone is repaired.
In this patch, a new table option tombstone_gc is added. The option is
used to configure tombstone gc mode. For example:
1) GC a tombstone after gc_grace_seconds
cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'timeout'} ;
This is the default mode. If no tombstone_gc option is specified by the
user. The old gc_grace_seconds based gc will be used.
2) Never GC a tombstone
cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'disabled'};
3) GC a tombstone immediately
cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'immediate'};
4) GC a tombstone after repair
cqlsh> ALTER TABLE ks.cf WITH tombstone_gc = {'mode':'repair'};
In addition to the 'mode' option, another option 'propagation_delay_in_seconds'
is added. It defines the max time a write could possibly delay before it
eventually arrives at a node.
A new gossip feature TOMBSTONE_GC_OPTIONS is added. The new tombstone_gc
option can only be used after the whole cluster supports the new
feature. A mixed cluster works with no problem.
Tests: compaction_test.py, ninja test
Fixes#3560
[avi: resolve conflicts vs data_dictionary]
In the next patches we will refactor mutation::consume(). Before doing
that add another test, which rebuilds the consumed mutation, comparing
it with the original.
This change makes row cache support reverse reads natively so that reversing wrappers are not needed when reading from cache and thus the read can be executed efficiently, with similar cost as the forward-order read.
The database is serving reverse reads from cache by default after this. Before, it was bypassing cache by default after 703aed3277.
Refs: #1413
Tests:
- unit [dev]
- manual query with build/dev/scylla and cache tracing on
Closes#9454
* github.com:scylladb/scylla:
tests: row_cache: Extend test_concurrent_reads_and_eviction to run reverse queries
row_cache: partition_snapshot_row_cursor: Print more details about the current version vector
row_cache: Improve trace-level logging
config: Use cache for reversed reads by default
config: Adjust reversed_reads_auto_bypass_cache description
row_cache: Support reverse reads natively
mvcc: partition_snapshot: Support slicing range tombstones in reverse
test: flat_mutation_reader_assertions: Consume expected range tombstones before end_of_partition
row_cache: Log produced range tombstones
test: Make produces_range_tombstone() report ck_ranges
tests: lib: random_mutation_generator: Extract make_random_range_tombstone()
partition_snapshot_row_cursor: Support reverse iteration
utils: immutable-collection: Make movable
intrusive_btree: Make default-initialized iterator cast to false
"
The second patch in this series is a mechanical conversion of
reader_concurrency_semaphore to flat_mutation_reader_v2, and caller
updates.
The first patch is needed to pass the test suite, since without it a
real reader version conversion would happen on every entry to and exit
from reader_concurrency_semaphore, which is stressful (for example:
mutation_reader_test.test_multishard_streaming_reader reaches 8191
conversions for a couple of readers, which somehow causes it to catch
SIGSEGV in diverse and seemingly-random places).
Note that in a real workload it is unreasonable to expect readers being
parked in a reader_concurrency_semaphore to be pristine, so
short-circuiting their version conversions will be impossible and this
workaround will not really help.
"
* tag 'rcs-v2-v4' of https://github.com/cmm/scylla:
reader_concurrency_semaphore: convert to flat_mutation_reader_v2
short-circuit flat mutation reader upgrades and downgrades
Make sure that major will compact data in all sstables and memtable,
as tombstones sitting in memtable could shadow data in sstables.
For example, a tombstone in memtable deleting a large partition could
be missed in major, so space wouldn't be saved as expected.
Additionally, write amplification is reduced as data in memtable
won't have to travel through tiers once flushed.
Fixes#9514.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Message-Id: <20211217160055.96693-2-raphaelsc@scylladb.com>
"
TWCS perform STCS on a window as long as it's the most recent one.
From there on, TWCS will compact all files in the past window into
a single file. With some moderate write load, it could happen that
there's still some compaction activity in that past window, meaning
that per-window major may miss some files being currently compacted.
As a result, a past window may contain more than 1 file after all
compaction activity is done on its behalf, which may increase read
amplification. To avoid that, TWCS will now make sure that per-window
major is serialized, to make sure no files are missed.
Fixes#9553.
tests: unit(dev).
"
* 'fix_twcs_per_window_major_v3' of https://github.com/raphaelsc/scylla:
TWCS: Make sure major on past window is done on all its sstables
TWCS: remove needless param for STCS options
TWCS: kill unused param in newest_bucket()
compaction: Implement strategy control and wire it
compaction: Add interface to control strategy behavior.
"
Users are adjusted by sprinkling `upgrade_to_v2()` and
`downgrade_to_v1()` where necessary (or removing any of these where
possible). No attempt was made to optimize and reduce the amount of
v1<->v2 conversions. This is left for follow-up patches to keep this set
small.
The combined reader is composed of 3 layers:
1. fragment producer - pop fragments from readers, return them in batches
(each fragment in a batch having the same type and pos).
2. fragment merger - merge fragment batches into single fragments
3. reader implementation glue-code
Converting layers (1) and (3) was mostly mechanical. The logic of
merging range tombstone changes is implemented at layer (2), so the two
different producer (layer 1) implementations we have share this logic.
Tests: unit(dev)
"
* 'combined-reader-v2/v4' of https://github.com/denesb/scylla:
test/boost/mutation_reader_test: add test_combined_reader_range_tombstone_change_merging
mutation_reader: convert make_clustering_combined_reader() to v2
mutation_reader: convert position_reader_queue to v2
mutation_reader: convert make_combined_reader() overloads to v2
mutation_reader: combined_reader: convert reader_selector to v2
mutation_reader: convert combined reader to v2
mutation_reader: combined_reader: attach stream_id to mutation_fragments
flat_mutation_reader_v2: add v2 version of empty reader
test/boost/mutation_reader_test: clustering_combined_reader_mutation_source_test: fix end bound calculation
Some implementation notes below.
When iterating in reverse, _last_row is after the current entry
(_next_row) in table schema order, not before like in the forward
mode.
Since there is no dummy row before all entries, reverse iteration must
be now prepared for the fact that advancing _next_row may land not
pointing at any row. The partition_snapshot_row_cursor maintains
continuity() correctly in this case, and positions the cursor before
all rows, so most of the code works unchanged. The only excpetion is
in move_to_next_entry(), which now cannot assume that failure to
advance to an entry means it can end a read.
maybe_drop_last_entry() is not implemented in reverse mode, which may
expose reverse-only workload to the problem of accumulating dummy
entries.
ensure_population_lower_bound() was not updating _last_row after
inserting the entry in latets version. This was not a problem for
forward reads because they do not modify the row in the partition
snapshot represented by _last_row. They only need the row to be there
in the latest version after the call. It's different for reveresed
reads, which change the continuity of the entry represented by
_last_row, hence _last_row needs to have the iterator updated to point
to the entry from the latest version, otherwise we'd set the
continuity of the previous version entry which would corrupt the
continuity.
Currently the test assumes that fragments represent weakly monotonic
upper bounds and therefore unconditionally overwrites the upper-bound on
receiving each fragment. Range tombstones however violate this as a
range tombstone with a smaller position (lower bound) may have a higher
upper bound than some or all fragments that follow it in the stream.
This causes test failures after the converting the combined reader to
v2, but not before, no idea why.
Appending an empty range adjacent to an existing range tombstone would
not deoverlap (by dropping the empty range tombstone) resulting in
different (non canoncial) result depending on the order of appending.
Suppose that range tombstone [a, b] covers range tombstone [x, x), and [a, x) and [x, b) are range tombstones which correspond to [a, b] split around position x.
Appending [a, x) then [x, b) then [x, x) would give [a, b)
Appending [a, x) then [x, x) then [x, b) would give [a, x), [x, x), [x, b)
The fix is to drop empty range tombstones in range_tombstone_list so that the result is canonical.
Fixes#9661Closes#9764
* github.com:scylladb/scylla:
range_tombstone_list: Deoverlap adjacent empty ranges
range_tombstone_list: Convert to work in terms of position_in_partition
Stop using database (and including database.hh) for schema related
purposes and use data_dictionary instead.
data_dictionary::database::real_database() is called from several
places, for these reasons:
- calling yet-to-be-converted code
- callers with a legitimate need to access data (e.g. system_keyspace)
but with the ::database accessor removed from query_processor.
We'll need to find another way to supply system_keyspace with
data access.
- to gain access to the wasm engine for testing whether used
defined functions compile. We'll have to find another way to
do this as well.
The change is a straightforward replacement. One case in
modification_statement had to change a capture, but everything else
was just a search-and-replace.
Some files that lost "database.hh" gained "mutation.hh", which they
previously had access to through "database.hh".
"
These two readers are crucial for writing tests for any composable
reader so we need v2 versions of them before we can convert and test the
combined reader (for example). As these two readers are often used in
situations where the payload they deliver is specially crafted for the
test at hand, we keep their v1 versions too to avoid conversion meddling
with the tests.
Tests: unit(dev)
"
* 'forwarding-and-fragment-reader-v2/v1' of https://github.com/denesb/scylla:
flat_mutation_reader_v2: add make_flat_mutation_reader_from_fragments()
test/lib/mutation_source_test: don't force v1 reader in reverse run
mutation_source: add native_version() getter
flat_mutation_reader_v2: add make_forwardable()
position_in_partition: add after_key(position_in_partition_view)
flat_mutation_reader: make_forwardable(): fix indentation
flat_mutation_reader: make_forwardable(): coroutinize reader
The test in its current form is invalid, as database::remove
does removing the table's name from its listing
as well as from the keyspace metadata, so it won't be found
after that.
That said, database::drop_column_family then proceeds
to truncate and stop the table, after calling await_pending_ops,
and the latter should indeed block on the lock taken by the test.
This change modifies the test to create some sstables in the
table's directory before starting the sstable_directory.
Then, when executing "drop table" in the background,
wait until the table is not found by db.find_column_family
That would fail the test before this change.
See https://jenkins.scylladb.com/job/scylla-enterprise/job/next/1442/artifact/testlog/x86_64_debug/sstable_directory_test.sstable_directory_test_table_lock_works.4720.log
```
INFO 2021-12-13 14:00:17,298 [shard 0] schema_tables - Dropping ks.cf id=00487bc0-5c1d-11ec-9e3b-a44f824027ae version=b10c4994-31c7-3f5a-9591-7fedb0273c82
test/boost/sstable_directory_test.cc(453): fatal error: in "sstable_directory_test_table_lock_works": unexpected exception thrown by table_ok.get()
```
A this point, the test verifies again that the sstables are still on
disk (and no truncate happened), and only after drop completed,
the table should not exist on disk.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Message-Id: <20211214104407.2225080-1-bhalevy@scylladb.com>
The series is on top of "wire up schema raft state machine". It will
apply without, but will not work obviously (when raft is disabled it
does nothing anyway).
This series does not provide any linearisability just yet though. It
only uses raft as a means to distribute schema mutations. To achieve
linearisability more work is needed. We need to at lease make sure
that schema mutation use monotonically increasing timestamps and,
since schema altering statement are RMW, no modification to schema
were done between schema mutation creation and application. If there
were an operation needs to be restarted.
* scylla-dev/gleb/raft-schema-v5: (59 commits)
cql3: cleanup mutation creation code in ALTER TYPE
cql3: use migration_manager::schema_read_barrier() before accessing a schema in altering statements
cql3: bounce schema altering statement to shard 0
migration_manager: add is_raft_enabled() to check if raft is enabled on a cluster
migration_manager: add schema_read_barrier() function
migration_manager: make announce() raft aware
migration_manager: co-routinize announce() function
migration_manager: pass raft_gr to the migration manager
migration_manager: drop view_ptr array from announce_column_family_update()
mm: drop unused announce_ methods
cql3: drop schema_altering_statement::announce_migration()
cql3: drop has_prepare_schema_mutations() from schema altering statement
cql3: drop announce_migration() usage from schema_altering_statement
cql3: move DROP AGGREGATE statement to prepare_schema_mutations() api
migration_manager: add prepare_aggregate_drop_announcement() function
cql3: move DROP FUNCTION statement to prepare_schema_mutations() api
migration_manager: add prepare_function_drop_announcement() function
cql3: move CREATE AGGREGATE statement to prepare_schema_mutations() api
migration_manager: add prepare_new_aggregate_announcement() function
cql3: move CREATE FUNCTION statement to prepare_schema_mutations() api
...
Appending an empty range adjacent to an existing range tombstone would
not deoverlap (by dropping the empty range tombstone) resulting in
different (non canoncial) result depending on the order of appending.
Suppose that [a, b] covers [x, x)
Appending [a, x) then [x, b) then [x, x) would give [a, b)
Appending [a, x) then [x, x) then [x, b) would give [a, x), [x, x), [x, b)
Fix by dropping empty range tombstones.
Once current window is sealed, TWCS is supposed to compact all its
sstables into one. If there's ongoing compaction, it can happen that
sstables are missed and therefore past windows will contain more than
one sstable. Additionally, it could happen that major doesn't happen
at all if under heavy load. All these problems are fixed by serializing
major on past window and also postponing it if manager refuses to run
the job now.
Fixes#9553.
Reviewed-by: Benny Halevy <bhalevy@scylladb.com>
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
STCS option can be retrieved from class member, as newest_bucket()
is no longer a static function. let's get rid of it.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
"
Today, data from different buckets (e.g. windows) cannot be compacted together because
mutation compactor happens inside each consumer, where each consumer is done on behalf
of a particular bucket. To solve this problem, mutation compaction process is being
moved from consumer into producer, such that interposer consumer, which is responsible
for segregation, will be feeded with compacted data and forward it into the owner bucket.
Fixes#9662.
tests: unit(debug).
"
* 'compact_across_buckets_v2' of github.com:raphaelsc/scylla:
tests: sstable_compaction_test: add test_twcs_compaction_across_buckets
compaction: Move mutation compaction into producer for TWCS
compaction: make enable_garbage_collected_sstable_writer() more precise
Verify that TWCS compaction can now compact data across time windows,
like a tombstone which will cause all shadowed data to be purged
once they're all compacted together.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
The main difference compared to v1 (apart from having _v2 suffix at
relevant places) is how slicing and reversing works. The v2 variant has
native reverse support built-in because the reversing reader is not
something we want to convert to v2.
A native v2 mutation-source test is also added.
do_with_some_data runs a function in a seastar thread.
It needs to get() the future func returns rather
than propagating it.
This solves a secondary failure due to abandoned future
when the test case fails, as seen in
https://jenkins.scylladb.com/view/master/job/scylla-master/job/next/4254/artifact/testlog/x86_64_debug/database_test.snapshot_with_quarantine_works.381.log
```
test/boost/database_test.cc(903): fatal error: in "snapshot_with_quarantine_works": critical check expected.empty() has failed
WARN 2021-12-08 00:35:16,300 [shard 0] seastar - Exceptional future ignored: boost::execution_aborted, backtrace: 0x10935e50 0x16ff2d8d 0x16ff2a4d 0x16ff5033 0x16ff5ec2 0x162d4ce9 0x10a2bdb5 0x10a2bd24 0x10a54ca4 0x10a27cf3 0x10a22151 0x10a67c9d 0x10a67a78 0x163ac37e 0x163b29e9 0x163b7690 0x163b51c1 0x17c212df 0x17c1f097 0x17bf8b4c 0x17bf83f2 0x17bf82a2 0x17bf7d52 0x10f8bf5a 0x166db84b /lib64/libpthread.so.0+0x9298 /lib64/libc.so.6+0x100352
...
*** 1 abandoned failed future(s) detected
Failing the test because fail was requested by --fail-on-abandoned-failed-futures
```
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
Message-Id: <20211209174512.851945-1-bhalevy@scylladb.com>
Mutations are not guaranteed to come in the order of their timestamps.
If there is an expired tombstone in the sstable and a repair inserts old
data into memtable, the compaction would not consider memtable data and
purge the tombstone leading to data resurrection. The solution is to
disallow purging tombstones newer than min memtable timestamp.
Since sstable reader was already converted to flat_mutation_reader_v2, compaction layer can naturally be converted too.
There are many dependencies that use v1. Those strictly needed like readers in sstable set, which links compaction to sstable reader, were converted to v2 in this series. For those that aren't essential we're relying on V1<-->V2 adaptors, and conversion work on them will be postponed. Those being postponed are: scrub specialized reader (needs a validator for mutation_fragment_v2), interposer consumer, combined reader which is used by incremental selector. incremental selector itself was converted to v2.
tests: unit(debug).
Closes#9725
* github.com:scylladb/scylla:
compaction: update compaction::make_sstable_reader() to flat_mutation_reader_v2
sstable_set: update make_crawling_reader() to flat_mutation_reader_v2
sstable_set: update make_range_sstable_reader() to flat_mutation_reader_v2
sstable_set: update make_local_shard_sstable_reader() to flat_mutation_reader_v2
sstable_set: update incremental_reader_selector to flat_mutation_reader_v2
"
Currently stateful (readers being saved and resumed on page boundaries)
multi-range scans are broken in multiple ways. Trying to use them can
result in anything from use-after-free (#6716) or getting corrupt data
(#9718). Luckily no-one is doing such queries today, but this started to
change recently as code such as Alternator TTL and distributed
aggregate reads started using this.
This series fixes both problems and adds a unit test too exercising this
previously completely unused code-path.
Fixes: #6716Fixes: #9718
Tests: unit(dev, release, debug)
"
* 'fix-stateful-multi-range-scans/v1' of https://github.com/denesb/scylla:
test/boost/multishard_mutation_query_test: add multi-range test
test/boost/multishard_mutation_query_test: add multi-range support
multishard_mutation_query: don't drop data during stateful multi-range reads
multishard_combining_reader: reader_lifecycle_policy: allow saving read range on fast-forward
As part of the drive to move over to flat_mutation_reader_v2, update
make_filtering_reader(). Since it doesn't examine range tombstones
(only the partition_start, to filter the key) the entire patch
is just glue code upgrading and downgrading users in the pipeline
(or removing a conversion, in one case).
Test: unit (dev)
Closes#9723
For each quarantine mode:
Validate sstables to quarantine one of them
and then scrub with the given quarantine mode
and verify the output whwther the quarantined
sstable was scrubbed or not.
Signed-off-by: Benny Halevy <bhalevy@scylladb.com>
