mirrors/scylladb
1
0
Fork 0
mirror of https://github.com/scylladb/scylladb.git synced 2026-05-02 14:15:46 +00:00
Code Issues Packages Projects Releases Wiki Activity
7,158 Commits 241 Branches 541 Tags
cc6d4e0fc98f98ff6f3bbdb252cf93050c5b2912
Commit Graph

7158 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Asias He
cc6d4e0fc9 storage_service: Introduce run_with_no_api_lock 
Run on cpu zero without take the api lock.
 2015-10-29 09:50:03 +08:00
Asias He
2c8867c348 config: Enable storage_port option 2015-10-29 08:58:41 +08:00
Asias He
97d2bdf663 storage_service: Serialize gossiping operations 2015-10-29 08:58:41 +08:00
Asias He
f2e5e318d7 storage_service: Serialize effective_ownershipp with read api lock 2015-10-29 08:58:41 +08:00
Asias He
94a49121c2 storage_service: Serialize get_ownership with read api lock 2015-10-29 08:58:41 +08:00
Asias He
d41cd48a5f storage_service: Serialize is_joined with read api lock 2015-10-29 08:58:41 +08:00
Asias He
24a452133a storage_service: Serialize join_ring with write api lock 2015-10-29 08:58:41 +08:00
Asias He
08337b0df1 storage_service: Serialize more node management operations with read api lock 
- get_operation_mode
- is_starting
- is_rpc_server_running
- is_native_transport_running
- get_load_map
- is_initialized
 2015-10-29 08:58:41 +08:00
Asias He
c6bf157610 storage_service: Introduce run_with_read_api_lock 
Similar to run_with_write_api_lock, but the operations only query the
cluster.
 2015-10-29 08:58:41 +08:00
Asias He
93154e0184 storage_service: Serialize more node management operations with write api lock 
- start_rpc_server
- stop_rpc_server
- start_native_transport
- stop_native_transport
- remove_node
 2015-10-29 08:57:45 +08:00
Raphael S. Carvalho
c2a98807c7 compaction_manager: fix remove 
remove() is the function used to remove every reference to a cf from
the compaction manager. This function works by removing cf from the
queue, and waiting for possible ongoing compaction on cf.
However, a cf may be re-queued by compaction manager task if there
is pending compaction by the end of compaction.

If cf is still referenced by the time remove() returns, we could end
up with an use-after-free. To fix that, a task shouldn't re-queue a
cf if it was asked to stop. The stat pending_tasks was also not
being updated when a cf was removed from the task queue.

Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
 2015-10-28 17:35:26 +02:00
Gleb Natapov
ac5f92db70 storage_proxy: clean up local_dc checking 
The only place local_dc is checked during mutation sending is in
send_to_live_endpoints(), but current code pass it there throw several
function call layers. Simplify the code by getting local_dc when it is
used directly.
 2015-10-28 16:10:18 +02:00
Paweł Dziepak
6e5916161c test/cql_test_env: merge client state after executing query 
Since 4641dfff24 "service: Copy client
state to query state" after executing a query client state needs to be
merged back. If that's not done client_state::_last_timestamp_micros
won't be advanced properly and mutations originating from the same
source may have exactly the same timestamp.

Signed-off-by: Paweł Dziepak <pdziepak@scylladb.com>
 2015-10-28 14:31:11 +02:00
Avi Kivity
4d23712c9c Merge "Fix gossip test" from Asias 2015-10-28 12:59:19 +02:00
Asias He
25c898fe9c gossip: Enable too more log prints for debug 2015-10-28 16:13:57 +08:00
Asias He
e0e8e9a1ed tests: Remove redundant debug info for gossip 
The debug info is printed in logger already. Avoid to print it twice.
 2015-10-28 16:13:57 +08:00
Asias He
01ee5d002a failure_detector: Remove debug print in operator<< 2015-10-28 16:13:57 +08:00
Asias He
f205ae30c3 tests: Fix gossip 
- scylla/seastar/core/sharded.hh:439: Service& seastar::sharded<Service>::local()
  [with Service = locator::snitch_ptr]: Assertion `local_is_initialized()' failed.
- ./utils/fb_utilities.hh:74: static const inet_address utils::fb_utilities::get_broadcast_address():
  Assertion `broadcast_address()' failed.
 2015-10-28 16:13:57 +08:00
Takuya ASADA
5aab69e72f dist: do not remove build/ dir when scylla-server ubuntu package building 
Signed-off-by: Takuya ASADA <syuu@scylladb.com>
 2015-10-27 18:23:09 +02:00
Avi Kivity
7f88db8625 Merge "storage_service and gossip update" from Asias 2015-10-27 18:14:43 +02:00
Lucas Meneghel Rodrigues
a9a33d5a99 test.py: PEP8 Fixes 
Fix some PEP8 problems found in the tester code:

 * Wrong spacing around operators
 * Lines between class and function definitions
 * Fixed some of the larger than 80 column statements
 * Removed an unused import

Signed-off-by: Lucas Meneghel Rodrigues <lmr@scylladb.com>
 2015-10-27 17:54:54 +02:00
Tomasz Grabiec
491eae58a4 Merge branch 'gleb/read-repair' from seastar-dev.git 
From Gleb:

Now that we can calculate mutation diffs do so on digest mismatch and send
them out.
 2015-10-27 16:36:05 +01:00
Asias He
7e6e90dc52 storage_service: Serialize decommission 
Prevent two operations happen simultaneously.
 2015-10-27 21:48:38 +08:00
Asias He
4b75815306 storage_service: Introduce run_with_write_api_lock helper 
It is useful to run code on cpu zero with the api lock.
 2015-10-27 21:48:38 +08:00
Asias He
62228726a3 storage_service: Introduce a rwlock to serialize management operations 2015-10-27 21:48:38 +08:00
Asias He
1bbc1920d2 range_streamer: Start to use get_preferred_ip 
It is available now.
 2015-10-27 21:48:37 +08:00
Asias He
306bab9ead storage_service: Use get_preferred_ip 
Now that it is available, use it.
 2015-10-27 21:48:37 +08:00
Asias He
55b76a8963 api: No cpu zero trick for remove_node 
storage_service::remove_node is guaranteed to run on cpu zero only.
 2015-10-27 21:48:37 +08:00
Asias He
b3c7305d25 storage_service: Make remove_node runs on cpu 0 only 
We need to serialize nodetool operations to avoid two operations
happening simultaneously. Running on cpu 0 is one step toward this
goal.
 2015-10-27 21:48:37 +08:00
Asias He
6c6b1c4ba7 storage_service: Make decommission runs on cpu 0 only 
We need to serialize nodetool operations to avoid two operations
happening simultaneously. Running on cpu 0 is one step toward this goal.
 2015-10-27 21:48:37 +08:00
Asias He
49daba2599 storage_service: Do not ignore future in decommission 
gossiper::stop returns a future which we can not ignore.
 2015-10-27 21:48:37 +08:00
Asias He
00311817bd storage_service: Implement shutdown_client_servers 2015-10-27 21:48:37 +08:00
Asias He
83eb36796f storage_service: Kill FIXME for LoadBroadcaster.BROADCAST_INTERVAL 
It is available now.
 2015-10-27 21:48:37 +08:00
Asias He
1469cec5bf gossiper: Kill free function helper to get heart version and generation number 
They can only be executed on cpu 0. Make the gossiper member
functions for them to do so.
 2015-10-27 21:48:37 +08:00
Asias He
f573059698 gossiper: Kill free function helper for {unsafe_,}assassinate_endpoint 
They can only be executed on cpu 0. Make the gossiper member functions
for them to do so.
 2015-10-27 21:48:37 +08:00
Asias He
c5f377eb8b gossip: Simplify get_endpoint_downtime 
_unreachable_endpoints is replicated to call cores. No need to query
on core 0.
 2015-10-27 21:48:37 +08:00
Asias He
6f1db4fb72 gossip: Simplify get_unreachable_members 
_unreachable_endpoints is replicated to call cores. No need to query on
core 0.

This also fixes a bug in storage_proxy::truncate_blocking
which might access _unreachable_endpoints on non-zero cores.
 2015-10-27 21:48:37 +08:00
Asias He
a9f96d1f5a gossip: Replicate _unreachable_endpoints to all cores 2015-10-27 21:48:37 +08:00
Asias He
2439a2a982 gossip: Simplify get_live_members 
_live_endpoints is replicated to call cores. No need to query on core 0.
 2015-10-27 21:48:37 +08:00
Asias He
a28ba9cde8 api: Simplify get_tokens get_node_tokens and get_token_endpoint 
token_metadata is replicated to all cores. No need to query on core 0.
 2015-10-27 21:48:37 +08:00
Asias He
4e886f0399 storage_service: Implement is_rpc_server_running 2015-10-27 21:48:37 +08:00
Asias He
7f0634b429 storage_service: Implement stop_rpc_server 2015-10-27 21:48:37 +08:00
Asias He
ca66bea619 storage_service: Implement is_native_transport_running 2015-10-27 21:48:37 +08:00
Asias He
7bc49a1efe storage_service: Implement stop_native_transport 2015-10-27 21:48:37 +08:00
Asias He
8218ab7922 storage_service: Implement start_native_transport and start_rpc_server 
They are used for APIs. Share the code in main.cc as well.
 2015-10-27 21:48:37 +08:00
Lucas Meneghel Rodrigues
42c0acfc44 test.py: Return test output only if subprocess succeeded 
The current code will try to print the output of a
subprocess.Popen().communicate() call even if that
call raised an exception and that output is None.

Let's fix this problem by only printing the output
if it's not None.

Signed-off-by: Lucas Meneghel Rodrigues <lmr@scylladb.com>
 2015-10-27 15:17:05 +02:00
Asias He
3c47844e8c storage_service: Complete check_for_endpoint_collision 
Part of it was stubbed.
 2015-10-27 21:17:02 +08:00
Asias He
77a87cb2b6 storage_service: Implement prepare_replacement_info 
Needed by replace node operation.
 2015-10-27 21:17:02 +08:00
Vlad Zolotarov
5cdbc3701a tests: set broadcast address 
Since commit 5613979a85
broadcast address has to be set before it's used for the first
time.

Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
 2015-10-27 15:16:13 +02:00
Avi Kivity
345f739f28 Merge seastar upstream 
* seastar 501e4cb...9d8913a (3):
  > Add mutable to with_lock and do_with
  > app-template: disable collectd by default
  > reactor: use fdatasync() instead of fsync()
 2015-10-27 15:13:36 +02:00