ce7a3a6912
Merge branch 'master' into no-disk-during-restore
4 years ago
0b3a41bd3f
Clearer comment
4 years ago
535d690d7c
Tigthen up Store-level locking
...
Previous locking didn't look quite correct, and with the advent of the
Serialize functionality, it needs to be more precise.
4 years ago
0d0bb3f44a
Close database before restoration
...
This change also adds better logging, so the restoration process can be
better observed.
4 years ago
eb01e8dc9d
Skip temp file when restoring in-memory database
...
This saves a read and write to disk, the size of the SQLite database.
4 years ago
db64f4c923
Don't use temp file during database snapshot
...
Snapshotting the database does not now need to write to disk, and copy
the SQLite database straight to RAM, for handoff to the Raft subsystem.
4 years ago
93a78f09a0
Add call to Serialize function
4 years ago
d560356d8d
Fix spelling mistakes
4 years ago
8576ed3cfd
Remove unused byte slice address
4 years ago
bb31aef20a
ABL
4 years ago
9b0852e711
Implement policy over trailing logs
...
Don't allow control over trailing logs directly, just implement a policy
that scales the number by 1.25 relative to Snapshot threshold. Without
doing this it may be that setting the snapshot threshold lower doesn't
actually affect SQLite start up times, due to log application. Perhaps
this really needs exposure at the command line, but the command line is
starting to get complicated.
4 years ago
8cb88d7744
Count number of legacy commands unmarshaled
4 years ago
4f0c3092b3
Better command decoding panic message
4 years ago
7780a0987b
Simplify variable declaration
4 years ago
e52b4c58c3
Factor out binary value functions
4 years ago
011bc00784
Add SQLite database page-centric size to status output
4 years ago
6d77d904e7
Compress SQLite database in Raft snapshot
...
Before Raft truncates the log, it needs to retrieve a snapshot of the SQLite database to represent the log entries that are removed during that truncation. Raft then writes that snapshot to disk. To reduce both disk IO and disk space this change compresses that SQLite snapshot before it is written to disk, and decompresses it during a restore operation.
This change also deals with older SQLite snapshots, which were not compressed.
4 years ago
d5322a1909
Ignore join request if node already part of cluster ( #711 )
...
rqlite used to work like this, but suffered a regression due to a change in how Hashicorp Raft worked. The manner it changed in was not public, so relying on it was always fragile.
4 years ago
6575de779d
Use Protobuf for encoding Raft Log commands
...
This PR changes Raft Log Entry encoding from JSON to Protobuf. Furthermore, larger Raft commands (which can result from batching SQL statements, or individually long SQL statements) are compressed before encoding.
This primary reason for this change is to reduce IO load since that is one of the largest performance bottlenecks. It will also reduce internode traffic.
Legacy JSON-encoded commands are still handled by this code, so this change is backwards-compatible with previous releases in the v5 series.
4 years ago
b4867713e4
Better variable name, Query -> SQL
4 years ago
b6ebdf3ab1
Support setting Leader Lease timeout
4 years ago
8a4a5c4901
Improve comments re opening of Store
4 years ago
f961bb4e44
Fix typos and comments
4 years ago
1480d8898f
Handle Raft commands missing Parameterized values
...
This allows this code to run with Raft logs from previous versions.
4 years ago
ac862e4091
Add system test for parameterized statements
...
Fix some bugs along the way too.
4 years ago
212d4e5c86
Store and Raft layer support parameterized queires
4 years ago
d840adc2a9
Convert Store to use new DB Statements
4 years ago
f81ef9024e
Add Raft log size on disk to status
4 years ago
c4a89ede8d
Add helper function for Raft log size
5 years ago
487c0e3f8c
Set Snapshot Interval to non-zero value
...
Snapshotting was *never* functional due to the zero interval value.
5 years ago
9c4151c37b
move comment down
...
Change-Id: I2f8c435adab5e1c46da2e6b91205ad4df850decc
5 years ago
9c295ca704
check level before freshness
...
Change-Id: I6d736ad09eac9f003f087adaa9c1f51691fd5c69
5 years ago
536cac4fdf
Allow log level control for Raft module
...
Default log level is INFO.
5 years ago
abb7772105
Support leader freshness with None consistency
...
This change allows the read request to specify the maximum time the node
receiving the request may have last heard from the cluster leader. It
only applies to a read consistency selection of None.
5 years ago
81f5f6faf1
Add support for non-voting nodes
...
A non-voting node doesn't participate in Raft consensus, but does
subscribe to the committed log entries originating with the leader.
This means a non-voting node keeps up-to-date with the state machine,
without impacting write-latency. These non-voting nodes can provide
read scalability for the cluster.
5 years ago
9d595e2ead
ABL
5 years ago
e0cf01cefc
Protect against possible nil pointer
5 years ago
a23af88235
Write backup and dump files to a writer
...
This avoids making an extra copy in RAM. Port of original PR at
https://github.com/rqlite/rqlite/pull/461/ .
5 years ago
f57ace7da2
Broadcast Store meta via standard consensus
...
With this change the cluster metadata (arbitrary key-value data associated with each node) is now broadcast across the cluster using the standard consensus mechanism. Specifically the use case for this metadata is to allow all nodes know the HTTP API address of all other nodes, for the purpose of redirecting requests to the leader.
This change removed the need for multiplexing two logical connections
over the single Raft TCP connection, which greatly simplifies the
networking code generally.
Original PR https://github.com/rqlite/rqlite/pull/434
5 years ago
e376812907
Add method to Store that returns leader ID
...
With the advent of Raft IDs, this distinction matters.
5 years ago
8a0e1ce8f0
Move to Hashicorp Raft v1
5 years ago
7352c376c7
Don't ignore Join error, actually return it
...
Clearly a typo.
5 years ago
085d2d23d8
ABL
5 years ago
0e69be3949
Store layer supports generating SQL format backups
...
Port PR453.
5 years ago
0471877f26
Abort transaction if load operation fails
5 years ago
ccca1ecfa9
Fix DSN key in Store status output
5 years ago
d47808a26b
Add expvar stats to Store
5 years ago
2e91858e1e
Allow Raft election timeout to be set
6 years ago
6937a377ae
Allow control of shutdown-on-remove
...
Defaults to leaving Raft run if the node is removed.
6 years ago
efec4d4e42
Return redirect if remove on follower
...
Fixes issue #391 .
7 years ago
b5b2cb21db
Log Store directory path on startup
7 years ago
6bbdc88c1d
ABL
...
Always be linting.
7 years ago
2a1b6c65c3
New QueryRequest and ExecuteRequest types
...
The Query() and Execute() functions on the Store now take a complex type
that encapsulates the statements, and associated parameters. This will
make it easier to add more control parameters when making requests.
7 years ago
303be73dcb
Add missing Store params to diagnostics
7 years ago
2617962f5f
Typo
8 years ago
4cb5035172
remove store.readPeersJSON
8 years ago
aa5c946a40
Linting
8 years ago
851f486ff8
Make waiting for logs optional
8 years ago
059ab0757b
Allow Store to wait for initial logs to be applied
...
Better fix for issue #260 .
8 years ago
4b6c936d53
Fix comment typos
8 years ago
ae0d119d83
Allow Raft Apply timeout to be configurable
8 years ago
5ed6ef0ae8
Covert dumps to queries in HTTP service
...
There is no need for special logic at the Store level.
8 years ago
e6586f41f2
Use single load for dump processing
...
Writing a sufficiently sophisicated parser is too much work, and unlikely to be successful soon. Instead this change simply loads the entire dump as one command, and allows the underlying SQLite support to parse it correctly. This will definitely work, but since the load goes over Raft, it may hit limits with regards to network transfer sizes. Right now any limitations in that area are unknown.
Therefore this functionality remains somewhat experimental.
8 years ago
098578bda8
Clarify guarantees of some functions
8 years ago
c2ab24378c
Lint fixes
8 years ago
2fb99e5b20
Use Store logger everywhere in Store
8 years ago
dc2128d02a
add a raw snapshotting API
...
This allows us to grab a raw snapshot of the databse.
v2: s/RawDBSnapshot/Database, remove reference to raft from the comment.
v3: add a leader arg to Database to do a leader check
v4: fix doc string.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
8 years ago
326ca757d7
add an API for using custom raft.PeerStore/log.Logger
...
This is useful in case the server needs to store other metadata (e.g. auth
data) along side the peer list.
The logging bit is handy in case something has its own logging framework
that it wants to use.
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
8 years ago
7ac3450ab2
Suport observer deregistration
8 years ago
aa47089a32
expose raft's observer API via store
...
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
8 years ago
4cf094c6a4
store: use ErrNotLeader everywhere
...
Signed-off-by: Tycho Andersen <tycho.andersen@canonical.com>
8 years ago
4242a641b1
Fix typo in comment
8 years ago
c00f1b460b
Accept BEGIN and COMMIT on load()
8 years ago
39ebf50cb7
Remove explicit control of FK constraints
...
It's not clear whether it's the right thing to do, and FK constraints
can still be controlled via PRAGMA commands.
8 years ago
9ffa309f69
Stop database-level control of FK during load
8 years ago
ed304b7840
Remove useless comment
8 years ago
36368d6786
Start removing batch load
8 years ago
68793cfcbb
Use a transaction for loaded batch
...
Bump the batch size to 1000 too.
8 years ago
f44c08a530
Skip loading blank lines
8 years ago
9c180d8f43
SQL parser now strips newlines
8 years ago
1f5cfc65d8
Trim leading and trailing newlines
8 years ago
84f3be9c10
Add missing package
8 years ago
b87de293ce
Move \n ; trimming to SQL line parser
8 years ago
6ee9cdbbdf
Start integration of SQL line parser
8 years ago
3726d1bb20
Add DB conf to diagnostics
8 years ago
cb7686fa46
Diagnostics show actual FK constraint state
8 years ago
2daac8704b
Support batching during load
8 years ago
31560160a1
Check FK state before load
8 years ago
4c6ee1ad39
Return error code during load if necessary
8 years ago
6007790ae4
Break loading when done
8 years ago
3286bf064d
Start adding load-from-dump
8 years ago
729221bf57
join endpoint returns HTTP 301 if necessary
8 years ago
95cb4baf2a
More cyclo changes
8 years ago
7bdedf8cd9
Reduce cyclo complexity of opening store
8 years ago
c40fff4d46
Support controlling FK constraints at startup
...
This was possible previously, but would need to be set everytime on
startup via the API. This change allows it to set at startup AND enables
foreign constraint checking by default.
8 years ago
0b13a1321d
Simplify error handling for Update Peers
8 years ago
aa467cffcd
Fix misspelling in comment
8 years ago
2982058a81
Support setting Raft heartbeat timeout
8 years ago
8a05236768
Support configurable snapshot thresholds
8 years ago
b67e56196e
add cluster state and leader functions to store
8 years ago
83026e4612
Add Peers to status output
8 years ago
47539cf8e1
Add RemoveNode() to Store
8 years ago
ef18cd5b17
Expose Raft Peers() on Store
8 years ago
bf2047b45b
'go lint' fixes
8 years ago
8c0fee4b06
Include Raft dir in status
9 years ago
5380c4d7e4
Map Raft "not leader" error to Store error ( #112 )
9 years ago
c04ff92d7a
Fill our API peers lookups ( #110 )
9 years ago
fac48629bc
Start initial integration of mux ( #109 )
...
Integrate TCP mux with cluster and store
This change allows any node, including followers, to use the Raft log to make changes to a cluster-wide state.
9 years ago
5aefda9983
Move to new rqlite org
9 years ago
e8c92e088c
Don't join unless actually necessary
9 years ago
a02f44f089
Remove cluster-like code from Store
9 years ago
9164550aab
Remove debug print
9 years ago
bdeda47774
Initial implementation of Meta server
...
This should really be a distinct cluster package.
9 years ago
0a3454a675
Methods on store to set and get API peers
9 years ago
0bffa38319
Start implementing cluster meta comms
9 years ago
16e3125149
Add support for cluster meta
...
This change required that the cluster meta become part of the Raft
snapshot.
9 years ago
99ebaf9767
Add support for peersSub
9 years ago
ed56f77714
Add hierarchy to Raft commands
9 years ago
099c7c18ac
Enable TCP muxing layer for Raft commns
9 years ago
ef554e33a0
Update sqlite3 status output
9 years ago
91a552b7ec
Further store-layer updates
9 years ago
31cda32a38
'go fmt' fixes
9 years ago
1ea4edd515
Store layer adapted to new DB layer
9 years ago
c07c11f134
Fix trivial typo
9 years ago
7fd7a9a385
Finish multi-node cluster testing
9 years ago
5bd3ab574d
Idiomatic network layer function names
9 years ago
9693d7dc95
Obey 'timings' flag at database level
9 years ago
5f5a666a06
Move to "weak" and "strong" for consistency levels
9 years ago
1a16588450
Fix 'timings' JSON tag
9 years ago
ac92530a6d
Add sqlite file size to status
9 years ago
db1fb33fd3
Allow timing control
...
Fixes issue #75 .
9 years ago
8f674219eb
Re-enable timings
9 years ago
1649b83d14
Fix unexpected command 'typ'
9 years ago
11c18cd769
Service and store support consistency levels
...
Fix issue #5 .
9 years ago
20a05fd0a1
Refactor Raft command to support queries
9 years ago
7b628686bf
Add SQLite version to diagnostics
9 years ago
62c17291b4
More diagnostic information
9 years ago
e7e212411a
Ensure Raft dir exists
9 years ago
4524d014eb
Leader redirection almost complete
9 years ago
9d614b7c8d
Add explicit error for "no leader"
9 years ago
3f1a3b3777
Fix Raft snapshots of in-memory databases
...
The snapshot code assumed all databases were disk-based.
9 years ago
8cb852d1cf
Prettier store address in status
9 years ago
59052ec0b4
More status output
9 years ago
e88f97ed06
Support cluster-level leader checks for queries
9 years ago
ce0a2ce269
Usage and comment fixes
9 years ago
f1963221ca
Small fixes and formatting
9 years ago
cc73f0abbc
Default to leader check before backup
9 years ago
2e7ae59245
Complete multi-node replication test
9 years ago
b058458a0f
Start adding multi-node tests
9 years ago
347109e4bb
More unit tests of store
9 years ago
Philip O'Toole
wangfenjin
WanliTian
Tycho Andersen
Benjamen Keroack