e6ddf27072
Unit test sync between chunked loading and Snapshotting
10 months ago
59e928d058
Remove superfluous blank line
10 months ago
df76560168
Sync and report number of loads in progress
10 months ago
b0cd5c5a0c
Add more Load stats
10 months ago
e0f361509a
Handle aborted chunks at lowest level
10 months ago
dd6f386450
Refactor chunk loading
10 months ago
94898ce43e
Track loads in progress
10 months ago
44fa23b714
Use TrailingLogs = 0 to truncate log
10 months ago
264e9822eb
Standardize on chunk size
10 months ago
d726a4c780
Correct and simplify logic
10 months ago
a96112b41d
Snapshot log after one-command load
10 months ago
1f1ebf209f
Don't delete last log
10 months ago
97e9b14bdf
Fix race during testing
10 months ago
8adc11fb83
Snapshot in a goroutine
10 months ago
4b272d04f1
Better use of counters in testing
10 months ago
c39d3472bc
Check snapshot occurs at unit level
10 months ago
c93447e29f
Delete chunked load entries in log after loading
10 months ago
0dcc67f6f1
Perform full snapshot after chunked load
...
This must happen as chunked loading directly changes the SQLite file,
invalidating any existing Snapshots. Also, add missing unit tests for
chunked loading.
10 months ago
31aef91d17
Add FullNeeded() to Snapshot store
...
It may be necessary to override the FullNeeded definition, so wrap it.
10 months ago
8211af961d
Move some VACUUM checks to the Store
10 months ago
5c7996fd98
Support optional VACUUM on backup
10 months ago
a0bf10b021
Move VACUUM to DB layer
10 months ago
f26f15476e
Actually add Provider type
10 months ago
dfb8a60c17
Move Store Provide to a distinct type
10 months ago
3ae4255e62
Backup-related cleanups
10 months ago
4c31c480f5
Better name
10 months ago
5a8a2744ff
Support skipping verifying frame data
...
This halves the Compacting Scan time, and doesn't seem necessary since
rqlite is the only system writing the WAL files.
10 months ago
ee13427c67
Tigher logging
10 months ago
bd9e96ea02
Integrate Compacting WAL writer
10 months ago
d53a34f765
Don't log too much
10 months ago
ba98e88a7b
Update store.go
10 months ago
70d5e6de34
Check join and notify addresses are resolvable
10 months ago
66c1002c32
Remove obsolete metrics
11 months ago
b2663de630
Rename snapshot2 -> snapshot
11 months ago
c11d059551
Better function name
11 months ago
10e6f1b115
Clean up snapshot FDs
11 months ago
93566415e3
Just run bad test
11 months ago
47d877cc25
Don't close dst file until done
11 months ago
09712788c9
Store now builds with snapshot2
11 months ago
6a60108f0c
Start work on Restore and Recover
...
Builds, but still needs more work. It's referencing snapshot(1).
11 months ago
d34c35df31
No need to Store-level WAL selection
11 months ago
5e3d481a6e
Fill out methods on SnapshotStore v2
12 months ago
7c420407c2
WIP snapshot2
...
Close() on the Sink is done (first pass). What's needed next are all the
methods on the Store (Open, List, FullNeeded). I also need to decide
what getSnapshots should really do -- return names or a slice of meta?
12 months ago
974f939ffe
Add an interface between Store and Snapshot Store
...
This will make it easier to swap new implementations in.
1 year ago
be17f2a3cc
Correct variable assignment order
1 year ago
5b4ff9b8ce
Fix test code
1 year ago
53b432a391
More DB and Snapshot stats
1 year ago
59e6a23b02
Clean-up logging
1 year ago
2fff800273
Substantially shrink Test_MultiNodeClusterLargeQueuedWrites
1 year ago
378dd4b2a1
Simpler logging of first application
1 year ago
5ab7069aa0
Update store.go
1 year ago
a4f7431108
Log first time a log is applied
1 year ago
ad86a2989b
Use non-racy function for Leader ID and address
1 year ago
225b5346f5
Add basic stats for Snapshot store
1 year ago
1ffb44cd38
ABL
1 year ago
854fba0e66
Better way to detect previous close
1 year ago
d78d2f4d9d
Short circuit EOF from decompressor
...
The gzip NewReader() seems to hang sometimes when reopened with the old
reader. Therefore just record that we have closed, and never reopen it.
1 year ago
3eb4607a74
GoDoc
1 year ago
a49385e1d3
Add end-to-end compress-decompress test
1 year ago
52aa68497d
Better comment on disabling multistream
1 year ago
06a94e50bf
Raft snapshotting runs to 100%
...
Still failing later.
1 year ago
cb384a9df2
More test tweaks
1 year ago
2395f30ddf
Be sure we don't write footer twice
1 year ago
57eabce899
Protect against double-close of NodeTransport
1 year ago
365abac57c
Simple unit test of NewNodeTransport
1 year ago
fedbdb5478
Enable GZIP snapshot compression
...
Tests are failing.
1 year ago
195981868f
Typo
1 year ago
6efdf9f2b9
Add Decompressor
1 year ago
1fe81713c9
More Compressor tests
1 year ago
55b3aaa805
Add initial version of Gzipping reader
1 year ago
d2c5acc447
Wrap transport as passthru
1 year ago
47e9bbde49
Fix single-node upgrade test
1 year ago
cb989d09a2
Some changes from PR v2
1 year ago
1a24557c19
Merge remote-tracking branch 'origin' into wal-snapshot-store-v3
1 year ago
208fc20659
Fix more uses of random
1 year ago
54713b63f4
Merge remote-tracking branch 'origin' into wal-snapshot-store-v3
1 year ago
c51b8364e8
Fix import path
1 year ago
2288cfe2a9
random only needed by store
1 year ago
c1a8b4a13a
Use global, sync'ed random source
...
Almost unbelievably it appears tests on AppVeyor get the same value when
seeding the random generator. See https://ci.appveyor.com/project/otoolep/rqlite/builds/47853399
1 year ago
c6e214de1a
Remember store attributes
1 year ago
dda86be0d3
Better Store open-and-close logging
1 year ago
c44d020c11
Only attempt to checkpoint once
1 year ago
d6b9c5b39e
Close streams
1 year ago
ee8b23ba94
More logging
1 year ago
e8e0381622
Log when Store closes
1 year ago
a6134b02bc
Instrument the FSM snapshot process
1 year ago
76867a682f
Time FSM snapshot creation
1 year ago
ab6ca783e1
Checkpoint database during Recovery
1 year ago
a9e895bee2
Some cleanups
1 year ago
bd0f78391e
Remove unneeded FSM snapshot code
1 year ago
47476d5b58
Integrate new snapshoting with Store
...
- Snapshot
- Restore
- RecoverNode
1 year ago
5eb4e38cbc
Add comments and errors
1 year ago
e4d5ffb8a9
Flesh our Store check()
1 year ago
48ea4bbfd8
Use a single header
1 year ago
ea8dc3ed18
More test fix-up post removal of in-memory
1 year ago
6895525482
Rename some tests
1 year ago
aa72408bc5
Remove racy use of rand
1 year ago
07d49961d8
Remove unused functions
1 year ago
9e31a054aa
RecoverNode now uses on-disk database
1 year ago
88ce311b37
Remove all WAL-related files when loading new DB
1 year ago
b931e387be
First pass at removing in-memory support
1 year ago
23b8b6e5c9
Actually rename
1 year ago
2ac27c2ffc
Move Store tests back to single directory
1 year ago
65c6a56468
Remove the file
1 year ago
88d448851d
Blank out on-disk store tests
1 year ago
02b7ea7955
Close Store so temp storage can be cleaned up
1 year ago
e47c7a1b7b
Stress-test snapshotting
1 year ago
e497ee75f9
Fix typo
1 year ago
f06e547ab5
Less disk IO during load
1 year ago
f5517c0f66
Correct comment
1 year ago
9bdf048975
Better naming of temp backup file
1 year ago
b5cb889e1e
Auto-restore doesn't require ready() check
1 year ago
b7633143cc
Restore from file uses chunking
1 year ago
5ca5ce342e
Support internode LoadChunk()
1 year ago
1c8e52c5f7
Test_SingleNodeLoadBinaryFromReader passes
...
That's good, but much more extensive testing needed, and this needs to
replace the existing load path -- and then we move to 8.0.
1 year ago
93a752ae0e
Better variable names
1 year ago
1c6f691ff8
Integrate chunking - WIP
...
I need more sophisticated dechunker management.
1 year ago
13d47b24d8
Store can load from an io.Reader
...
This may be helpful for chunked-loads.
1 year ago
a40c43725f
More logic reversal
1 year ago
f81aad8fd3
Revert logic for clarity
1 year ago
21dff4a68c
Remove on-disk-startup control
...
With the move to WAL and "synchronous mode" to OFF, on-disk startup
times are very close to in-memory. There is no need for this control
anymore and it complicates the start-up code.
1 year ago
e845e0c243
Fix remaining expression test cases
1 year ago
f7ced46377
Count number of DB stats errors
1 year ago
36fa7f2afd
Detect DB stats error and log
1 year ago
7ce2c6b99a
Fix comment
1 year ago
7432bceca3
Since we're passed a ReadCloser, close it
1 year ago
4342650c33
Fix RecoverNode
1 year ago
ad73779285
Periodically record actually applied index
...
This ensures the system doesn't apply uncommitted log entries to the
FSM, if it runs the in-memory startup. There have been no reports of
this in the field -- it is an edge case -- but this removes any chance
of it.
1 year ago
abfe056b4c
Remove any -wal and -shm files on shutdown
...
If in WAL mode, open-and-close again to remove the -wal file. This is not strictly necessary, since any on-disk database files will be removed when rqlite next starts, but it leaves the directory containing the database file in a cleaner state.
1 year ago
6cd8d30749
Fix Store-level database-backup test
1 year ago
a6925414ef
Better comments
1 year ago
b2f28d1181
Better WAL control
1 year ago
2f7672d897
Enable WAL mode
1 year ago
585f9c9c05
Use Snapshot decoder in Store
1 year ago
58534d4815
Better name for V1 Snapshotter
1 year ago
c87931c45c
Factor Snapshot creation into own module
1 year ago
3f462eecfa
Remove all on-disk SQLite files on startup
1 year ago
7e1b8d12d7
Add full WAL support to DB layer
1 year ago
eed215c664
Only 1 Snapshot needs to be retained
1 year ago
b9253e6b41
Improve logging in Store
1 year ago
4a7ac2850f
rqlite doesn't (yet) support WAL mode SQLite data
1 year ago
92bc513242
Move FSMSnapshot to own source file
1 year ago
e4a0d96f23
Further refactoring
1 year ago
3dd37f8771
Move on-disk Store testing to distinct file
1 year ago
6c7c32ab4e
Remove racy test code
1 year ago
aed541b98e
Poll nodes
1 year ago
8a8e126061
Use some loops
1 year ago
ef42d28e36
Refactor and log more during test
...
https://ci.appveyor.com/project/otoolep/rqlite/builds/47107026 - this
looks racy, as the Leader had been detected, but the config not yet
applied. Leave for now and see if more info can be gathered. This may be
happening because WaitForLeader() has a fast path now.
1 year ago
1b299b7e9b
Tigher logic
1 year ago
9faea33da7
Give time for new leader to be elected
1 year ago
Philip O'Toole