Metadata is read using default priority class, which can significantly
slow down the process under high load. Compaction class can be used,
and if it turns out to be a problem, we can switch to a special class
for it.
Fixes#1859.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Message-Id: <20170517184546.17497-1-raphaelsc@scylladb.com>
SSTable load temporarily uses more space than needed to store metadata,
due to:
1) All components are read using read_simple() which uses 128k buffer.
file::dma_read_bulk() will allocate 128k, and may potentially allocate
another big buffer (128k - read) for file::read_maybe_eof().
2) read_filter() may use double the space it needs to.
Due to the fact that sstable loading parallelism is unlimited, Scylla
may require much more memory to load all sstables, and that may lead to
OOM. Higher the number of sstables higher the memory overhead.
To confirm this problem, I wrote a test[1] which loads 30k sstables in
parallel and reports the memory usage peak in the end.
When loading 30k sstables, each of which metadata is ~300kb, memory
usage peak was ~18G. When loading completed, only ~9GB were needed to
store all the metadata.
[1]: https://gist.github.com/raphaelsc/2db37b4fb34301833ab9eeed3b1a524d
To fix this problem, we need to set a limit on load parallelism (let's
start with a small number like 3 and adjust later if needed) and rely
on readahead so that the requirement drops considerably without
increasing boot time. Actually, boot time is improved by it.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Reviewed-by: Nadav Har'El <nyh@scylladb.com>
When generating updates for a materialized view we need to read the
existing base row, to be able to determine the primary key of the view
row the new base update will supplant, in case the view includes a
base non-primary key column in its own primary key. That old view row
will be tombstoned or updated, if it exists, depending on the difference
between the new base row and the existing one, if any.
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
"This patch series adds CQL front-end support for secondary indices. You
can now execute CREATE INDEX and DROP INDEX statements, which will
update the newly added "Indexes" system table. However, the indexes are
not actually backed up by anything nor are they available for CQL
queries. The feature is hidden behind a new cluster feature flag and
enabled only with the "--experimental" flag."
* 'penberg/cql-2i/v2' of github.com:cloudius-systems/seastar-dev: (34 commits)
schema: Kill index_type enum
schema: Kill index_info class
cql3/statements/create_index_statement: Use database::existing_index_names() in validation
cql3/statements: Use secondary index manager in alter_table_statement class
index: Add secondary_index_manager
thrift/handler: Use index_metadata
db/schema_tables: Index persistence
schema: Add all_indices() to schema class
schema: Remove add_default_index_names() from schema_builder class
db/schema_tables: Add system table for indices
cql3/Cgl.g: DROP INDEX
cql3/statements: Add drop_index_statement class
database: Add find_indexed_table() to database class
cql3: Return change event from announce_migration()
cql3/statements: Multiple index targets for CREATE INDEX
cql3/statements: Use index_metadata in create_index_statement class
cql3/statements: Use feature flag in create_index_statement class
service/storage_service: Add feature flag for secondary indices
database: Add get_available_index_name() to database class
schema: Add get_default_index_name() to index_metadata class
...
NOTE: it's not wired yet.
Currently, a shared sstable is rewritten at all shards it belongs
to and only after that, it's deleted. With this new algorithm, a
shared sstable will be read only once and N unshared sstables
will be created, each of them with 1/N of the data. After it's
done, each owner shard will receive its new unshared sstable
replacing its ancestors.
Another benefit is that we'll no longer have resharding resulting
in number of sstables growing considerably after resharding.
A full-sized leveled sstable is usually 160MB, so after resharding,
we could have N files of 160MB/N. Now, leveled strategy will help
resharding. N adjacent sstables of same level will be resharded
together, so we'll end up with N files of N*160MB/N.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
When resharding, we're working with sstables from all shards. So let's say
we're done with resharding of sstable A that belongs to shard 0 and 1 and
sstable B that belongs to shard 1 and 2. SStables were generated for
shards 0, 1, and 2. So shards 0, 1, and 2 need to load the new sstables
and remove the ancestors. Shard 1 for example will remove sstables A and
B (ancestors) and add the new one. Then it comes this new function.
We'll forward new sstables to their target shards using foreign sstable
open info.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
For new resharding, it's important to exclude resharding sstables
from the list of candidates for regular compaction. That's doesn't
affect current resharding because it marks the sstables as
compacting. That won't work with new resharding which will work
with sstables from multiple shards.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
That's gonna be useful to quickly determine if it's worth resharding
a column family.
Reviewed-by: Nadav Har'El <nyh@scylladb.com>
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
column_family constructor uses delegation, as such, only the actual
constructor implementation should contain a call to register the
metrics.
Current implementation ends up with re registration of the metrics.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
Message-Id: <20170320140817.22214-1-amnon@scylladb.com>
This patch ensures we upgrade the mutation to the current schema when
generating and pushing view updates, so that the it matches the most
up to date views.
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
The write path uses a base schema at a particular version, and we
want it to use the materialized views at the corresponding version.
To achieve this, we need to map the state currently in db::view::view
to a particular schema version, which this patch does by introducing
the view_info class to hold the state previously in db::view::view,
and by having a view schema directly point to it.
The changes in the patch are thus:
1) Introduce view_info to hold the extra view state;
2) Point to the view_info from the schema;
3) Make the functions in the now stateless db::view::view non-member;
4) Remove the db::view::view class.
All changes are structural and don't affect current behavior.
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
"This series adds various optimisations to counter implementation
(nothing extreme, mostly just avoiding unnecessary operations) as well
as some missing features such as tracing and dropping timed out queries.
Performance was tested using:
perf-simple-query -c4 --counters --duration 60
The following results are medians.
before after diff
write 18640.41 33156.81 +77.9%
read 58002.32 62733.93 +8.2%"
* tag 'pdziepak/optimise-counters/v3' of github.com:cloudius-systems/seastar-dev: (30 commits)
cell_locker: add metrics for lock acquisition
storage_proxy: count counter updates for which the node was a leader
storage_proxy: use counter-specific timeout for writes
storage_proxy: transform counter timeouts to mutation_write_timeout_exception
db: avoid allocations in do_apply_counter_update()
tests/counters: add test for apply reversability
counters: attempt to apply in place
atomic_cell: add COUNTER_IN_PLACE_REVERT flag
counters: add equality operators
counters: implement decrement operators for shard_iterator
counters: allow using both views and mutable_views
atomic_cell: introduce atomic_cell_mutable_view
managed_bytes: add cast to mutable_view
bytes: add bytes_mutable_view
utils: introduce mutable_view
db: add more tracing events for counter writes
db: propagate tracing state for counter writes
tests/cell_locker: add test for timing out lock acquisition
counter_cell_locker: allow setting timeouts
db: propagate timeout for counter writes
...
"Work on this series started with fixing the 'nodetool clearsnapshot'.
The current master code ignores the snapshots in deleted keyspaces (issue #2045).
I noticed that in many places our code has to build the path to some directory/file
it simply had the sstring(<path1>) + "/" + sstring(<path2>) constructs which may cause us issues
if somebody decides to complile/run scylla on not-Unix-based OS, like Microsoft Windows.
I understand that this is a long shot but if we can make it right now - why not to.
The answer is boost::filesystem::path class - its synchronous parts, of course.
I decided to take an initiative and fix the issues above and then use the fixed code for
fixing the issue #2045:
- Fix some minor issues in the existing code.
- Extend the lister class and move it into the separate files outside database.cc.
On the way I've found an issue in the existing code (issue #2071).
This series fixes this one too (PATCH2)."
In the later stages of counter write path a mutation is produced that
already has all cells transformed to counter shards and can be applied
to the memtable and written to the commitlog.
The current interface expectes a frozen mutation, which is suboptimal
for counters. The freeze itself is unaviodable -- it is required by
commitlog, but we can avoid later deserialization of frozen_mutation
when it is applied to the memtable if we pass the unfrozen mutation
along.
"This introduces an API which allows forward navigation in a stream of mutation
fragments. It allows one to consume only a subset of the stream by iteratively
specifying sub-ranges from which fragments should be returned.
API outline:
When in forwarding mode, the stream does not return all fragments right away,
but only those belonging to the current range. Initially current range only
covers the static row. The stream can be forwarded, even before reaching end-
of-stream for current range, to a later range with fast_forward_to().
Forwarding doesn't change initial restrictions of the stream, it can only be
used to skip over data.
Monotonicity of positions is preserved by forwarding. That is fragments
emitted after forwarding will have greater positions than any fragments
emitted before forwarding.
For any range, all range tombstones relevant for that range which are present
in the original stream will be emitted. Range tombstones emitted before
forwarding which overlap with the new range are not necessarily re-emitted.
When not in forwarding mode, the stream acts as if the current range was equal
to the full range. This implies that fast_forward_to() cannot be
used.
Whether stream is in forwarding mode or not is specified when the stream
is created, typically via mutation_source interface.
What's left for later series:
Optimization by providing specialized implementations. This series implements
forwarding support in all mutation sources via generic wrapper which simply
drops fragments."
* tag 'tgrabiec/clustering-fast-forward-to-v2' of github.com:scylladb/seastar-dev:
tests: mutation_source_tests: Verify monotonicty of positions
tests: random_mutation_generator: Spread the keys more
tests: mutation_source_test: Make blobs more easily distinguishable
tests: streamed_mutation: Test that merged stream passes mutation source tests
tests: mutation_source_test: Add tests for forwarding of streamed_mutation
tests: streamed_mutation_assertions: Add methods for navigating the stream
tests: Add range generators to random_mutation_generator
partition_slice_builder: Add with_ranges()
query: Introduce full_clustering_range
streamed_mutation: Add non-owning variant of mutation_from_streamed_mutation()
db: Enable creating forwardable readers via mutation_source
mutation_source: Document liveness requirements
mutation_source: Cleanup
db: Replace virtual_reader_type with mutation_source_opt
partition_version: Refactor make_partition_snapshot_reader() overloads
database: Fix mutation_source created by as_mutation_source() to not ignore trace_state_ptr
memtable: Accept all mutation_source parameters
streamed_mutation: Implement fast_forward_to() in stream merger
streamed_mutation: Add generic implementation of forwardable streamed_mutation
streamed_mutation: Add fast_forward_to() API
position_in_partition: Introduce position_range
position_in_partition: Introduce position constructor for right after the static row
streamed_mutation: Make cast to view non-explicit
streamed_mutation: Make schema() getter non-copying
It was using the state passed via as_mutation_source() instead. Let's
respect mutation_source contract instead, and use the state passed via
mutation_source invocation.
Technically just a cleanup. Alse prerequisite for more cleanup.
5a0955e89d "db: add operations for
applying counter updates" merged two column_family::apply() overloads
into do_apply() in order to reduce code duplication. Unfortunately,
a call to check_valid_rp() didn't survive that change.
Message-Id: <20170221133800.30411-1-pdziepak@scylladb.com>
Move lister class away from database.cc.
This is a preparation for moving it to the seastar library.
Signed-off-by: Vlad Zolotarov <vladz@scylladb.com>
The current implementation of 'nodetool clearsnapshot' command only deletes the snapshots
of the keyspaces that are alive at the time the command is issued (issue #2045).
This, besides not implementing the spec, prevents users from being able to clear
the disk space occupied by snapshots of deleted keyspaces that are no longer needed
(e.g. snapshots created when KS is deleted).
This patch fixes this issue by making the database::clear_snapshot() scan the data directories
looking for the snapshots to be deleted instead of relying on in-memory data structures.
This patch makes column_family::clear_snapshot() method not needed any more.
Fixes#2045
Signed-off-by: Vlad Zolotarov <vladz@scylladb.com>
Filter should get all information that the caller has in hand that may be used for filtering.
directory_entry has the following information:
- Type of the entry
- Its name
For the code that used lister filters so far this would be enough, however it's not
hard to imagine a filter that may need the parent directory as well.
We will add the parent directory path in the follow up patches to make the interface
complete.
Signed-off-by: Vlad Zolotarov <vladz@scylladb.com>
"This series uses the newly added histogram and label support to add metrics to
the storage_proxy and to the column_family.
This would add latency and histogram and the missing metrics from column family."
* 'amnon/histogram_metrics' of github.com:cloudius-systems/seastar-dev:
database: add metrics registration for the coloumn family
storage_proxy: add read and write latency histogram
estimated_histogram: returns a metrics histogram
This patch adds a metrics registration to the column_family.
Using label each column metrics is label with its keyspace and column
family name.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
This patch changes the database apply path so that it also
generates the mutations for the column family's views and
sends them to the paired view replicas.
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
This patch adds the generate_view_updates() function to the
column_family class, which will use the view_update_builder to
generate updates to the column_family's materialized views.
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
Instead of storing the view in the column_family's map of materialized
views, store a lw_shared_ptr so that the view can be removed while it
is being updated.
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
"Before, the logic for releasing writes blocked on dirty worked like this:
1) When region group size changes and it is not under pressure and there
are some requests blocked, then schedule request releasing task
2) request releasing task, if no pressure, runs one request and if there are
still blocked requests, schedules next request releasing task
If requests don't change the size of the region group, then either some request
executes or there is a request releasing task scheduled. The amount of scheduled
tasks is at most 1, there is a single releasing thread.
However, if requests themselves would change the size of the group, then each
such change would schedule yet another request releasing thread, growing the task
queue size by one.
The group size can also change when memory is reclaimed from the groups (e.g.
when contains sparse segments). Compaction may start many request releasing
threads due to group size updates.
Such behavior is detrimental for performance and stability if there are a lot
of blocked requests. This can happen on 1.5 even with modest concurrency
because timed out requests stay in the queue. This is less likely on 1.6 where
they are dropped from the queue.
The releasing of tasks may start to dominate over other processes in the
system. When the amount of scheduled tasks reaches 1000, polling stops and
server becomes unresponsive until all of the released requests are done, which
is either when they start to block on dirty memory again or run out of blocked
requests. It may take a while to reach pressure condition after memtable flush
if it brings virtual dirty much below the threshold, which is currently the
case for workloads with overwrites producing sparse regions.
I saw this happening in a write workload from issue #2021 where the number of
request releasing threads grew into thousands.
Fix by ensuring there is at most one request releasing thread at a time. There
will be one releasing fiber per region group which is woken up when pressure is
lifted. It executes blocked requests until pressure occurs."
* tag 'tgrabiec/lsa-single-threaded-releasing-v2' of github.com:cloudius-systems/seastar-dev:
tests: lsa: Add test for reclaimer starting and stopping
tests: lsa: Add request releasing stress test
lsa: Avoid avalanche releasing of requests
lsa: Move definitions to .cc
lsa: Simplify hard pressure notification management
lsa: Do not start or stop reclaiming on hard pressure
tests: lsa: Adjust to take into account that reclaimers are run synchronously
lsa: Document and annotate reclaimer notification callbacks
tests: lsa: Use with_timeout() in quiesce()
* seastar 397685c...c1dbd89 (13):
> lowres_clock: drop cache-line alignment for _timer
> net/packet: add missing include
> Merge "Adding histogram and description support" from Amnon
> reactor: Fix the error: cannot bind 'std::unique_ptr' lvalue to 'std::unique_ptr&&'
> Set the option '--server' of tests/tcp_sctp_client to be required
> core/memory: Remove superfluous assignment
> core/memory: Remove dead code
> core/reactor: Use logger instead of cerr
> fix inverted logic in overprovision parameter
> rpc: fix timeout checking condition
> rpc: use lowres_clock instead of high resolution one
> semaphore: make semaphore's clock configurable
> rpc: detect timedout outgoing packets earlier
Includes treewide change to accomodate rpc changing its timeout clock
to lowres_clock.
Includes fixup from Amnon:
collectd api should use the metrics getters
As part of a preperation of the change in the metrics layer, this change
the way the collectd api uses the metrics value to use the getters
instead of calling the member directly.
This will be important when the internal implementation will changed
from union to variant.
Signed-off-by: Amnon Heiman <amnon@scylladb.com>
Message-Id: <1485457657-17634-1-git-send-email-amnon@scylladb.com>
That's because a single shard is used to calculate generation for new
sstables in upload directory, and that will result in that single shard
sharing all the resources with other shards.
For refresh without upload dir, it currently works fine because we
reshuffle column family dir instead.
flush_upload_dir() is now a free function, takes a distributed database
object, and uses calculate_shard_from_sstable_generation() to decide
which shard will move sstable using its own generation namespace.
Fixes#2008.
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Message-Id: <b0cccf7bbb61416ff8718bac92fdca90cc5fb9c9.1484253232.git.raphaelsc@scylladb.com>
This patch ensures that when adding a shared sstable, we select only
one cpu to update that column family's stats. This is important so we
don't overestimated the on-disk size of sstables when resharding
This fixes only a temporary miscount of the current load, since shared
sstables are eventually re-written, but a fixes a permanent miscount
of the total load.
Refs #1592
Signed-off-by: Duarte Nunes <duarte@scylladb.com>
Message-Id: <20170119144823.31041-1-duarte@scylladb.com>
