Align with rest of file (for better or worse). This allows calls from
entity without query_processor handy (i.e. storage_proxy).
Added "minimal" setup method for the "global" state, to facilitate
tests. Doing a full setup either in cql_test_env or after it is created
breaks badly. (Not sure why). So quick workaround.
Updated the current two users (batchlog_manager and commitlog_replayer)
callsites to conform.
* seastar 5fe596a...283901a (10):
> Add filesystem "link_file"
> scripts: posix_net_conf.sh: take care of the case with more than 32 CPUs
> posix: Add explanatory string to throw_system_error_on()
> tests: fix memory leak in thread_test test_thread_1
> tests: fix memory leak in timertest cancellation test
> README: require xfsprogs-devel
> file: query dma alignment from OS
> file: separate disk and memory dma alignment
> scripts: posix_net_conf.sh: Exclude CPU0 in RDS config for EN.
> README: Add missing Ubuntu 14.04 dependencies to README.md
Although it is technically a seastar problem, most complains about that is
coming from the Scylla side. I prefer to keep the message here so we can reference
a Scylla issue.
Signed-off-by: Glauber Costa <glommer@scylladb.com>
When we write an SSTable, all its components are already in memory. load() is
to big of a hammer.
We still want to keep the write operation separated from the preparation to
read, but in the case of a newly written SSTable, all we need to do is to open
the index and data file.
Fixes#300
Signed-off-by: Glauber Costa <glommer@scylladb.com>
cql2_type is a simple wrapper around data_type. But some data_types
(collection_type_impl) contain a cql3_type as a cache to avoid recomputing
it, resulting in a circular reference. This leaks memory when as_cql3_type()
is called.
Fix by using a static hash table for the cache.
That's important to guarantee that all other components were
deleted before deleting TemporaryTOC.
Signed-off-by: Raphael S. Carvalho <raphaelsc@cloudius-systems.com>
For us, everything is "off heap", so this will just be the total amount of
memory used by the filters.
Fixes#339
Signed-off-by: Glauber Costa <glommer@scylladb.com>
Do it so we can estimate how much memory it is being used by the filters. This
estimate is not 100 % correct: the implementation of the bloom_filter class
uses a thread-local variable that is common to all filters. We won't include
that in the estimate. But aside from that, it should be quite accurate.
Signed-off-by: Glauber Costa <glommer@scylladb.com>
When start a cluster on AWS, the seed node might get ready after
non-seed nodes is ready to contact it. Wait for seed node longer to make
the boot up process more robust.
This is for testing virtio mode, since we don't officially recommend to use virtio mode we should drop it.
Signed-off-by: Takuya ASADA <syuu@cloudius-systems.com>
With this change, we can see the number and length of compaction
activity per shard from collectd.
Signed-off-by: Raphael S. Carvalho <raphaelsc@cloudius-systems.com>
WARN level is for messages which should draw log reader's attention,
journalctl highlights them for example. Populating of keyspace is a
fairly normal thing, so it should be logged on lower level.
When scylla is deployed on AWS's c4.8xlarge or c4.8xlarge large
instances, we can apply irq and network queue binding to achieve better
performance.
Also make scylla skip using cpu0 which will be busy serving network
interrupts under high workload.
Scylla needs thrift and antlr3 which are not provided by ubuntu. We need
to compile them from source in order to build scylla. For now, let's
only support build on Fedora.
