... by extracting the db/rp from the given path.
Now that the code has "standardized" on extracting db/rp this way, the
ShardLocation struct is no longer necessary and thus has been removed.
We're back on the previous style of passing the path and walPath to
NewShard.
The current go compiler at the tip of the go master (1d5001af) has a modified implementation of
testing.quick.Check that now generates nil slices as test data. (See: https://gophers.slack.com/archives/general/p14567053570110). The existing tests expect round tripping to work in this case
but it does not. So, in these cases we change the expectation to reflect actual behaviour.
This needs to be checked for reasonableness.
Previously, the for loop at the end of the method assumed that all entries
had been deduplicated, including the entry discovered in the snapshot.
However, this wasn't actually true. With this change, we make it true.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Consider the write sequence: 6,1,snapshot,7,2.
The hot cache gets deduplicated, so is 2,7.
Now consider the test if 1 >= 2, this is false, so needSort is not set to true.
The problem is the implicit assumption that the snapshot is always sorted
by the time that merged() runs, but this may not be true.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Previously, we needed a write lock on the cache because it was the
only lock we had available to guard updates to entry.values and
entry.needSort.
However, now we have a entry-scoped lock for this purpose, we don't
need the cache write lock for this purpose. Since merged() doesn't
modify the .store or the c.snapshot.sort, there is no need for
a write lock on the cache to protect the cache.
So, we don't need to escalate here - we simply rely on the entry lock
to protect the entries we are iterating over.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Based on @jwilder's alternative to the 'dirty' slice that featured
in previous iterations of this fix.
Suggested-by: Jason Wilder <jason@influxdb.com>
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Currently two compactors can execute Engine.WriteSnapshot at once.
This isn't thread safe since both threads want to make modifications to
Cache.snapshot at the same time.
This commit introduces a lock which is acquired during Snapshot() and
released during ClearSnapshot(), ensuring that at most one thread
executes within Engine.WriteSnapshot() at once.
To ensure that we always release this lock, but only release the
snapshot resources on a successful commit, we modify ClearSnapshot() to
accept a boolean which indicates whether the write was successful or not
and guarantee to call this function if Snapshot() has been called.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
There are two tests that show two different one vulnerability.
One test shows that Cache.Deduplicate modifies entries in a snapshot's
store without a lock while cache readers are deduplicating those same
entries while correctly locked.
A second test shows that two threads trying to execute the methods
that Engine.WriteSnapshot calls will cause concurrent, unsynchronized
mutating access to the snapshot's store and entries.
The tests fail at this commit and are fixed by subsequent commits.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Fix for #5804.
The commit for #5789 rendered the semantics of snapshotCount statistic
useless. This commit restores semantics that have diagnostic value to
this statistic.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
The Cache had support for taking multiple snapshots to support writing
multiple snapshots to TSM files concurrently if that happened to be
a bottleneck. In practice, this is never a bottleneck and we only
run one snappshoting goroutine continously per shard which has worked
well for all workloads.
The multiple snapshot support introduces some unhandled failure scenarios
where wal segments could be removed without writing them to TSM files. If
a snapshot compaction fails to write due to transient disk errors, subsequent
snapshots will continue, but the failed one will not be retried. When the
subsequent ones succeeded, all closed wal segments are removed causing data
loss.
This change simplifies the snapshotting capability to ensure that there is only
ever one snapshot. If one fails, the next snapshot will update the existing
snapshot and retry all of old and new data.
Fixes#5686
The cache had some incorrect logic for determine when a series needed
to be deduplicated. The logic was checking for unsorted points and
not considering duplicate points. This would manifest itself as many
points (duplicate) points being returned from the cache and after a
snapshot compaction run, the points would disappear because snapshot
compaction always deduplicates and sorts the points.
Added a test that reproduces the issue.
Fixes#5719
The intent of this change is to avoid writing caches created for
snapshot cache instances into the tsm1_cache measurement. We can do
this by avoiding use of the NewCache constructor. All other methods
are only intended to be called from on the engine cache - never
on a snapshot.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Since we are not locking but relying on atomic arithmetic,
use Add rather than Set. Will also result in slightly less garbage
being created.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
The intent of this change is to ensure that all statistic fields of the
resulting tsm1_cache measurement are initialized on initialization of
the cache. That way, any consumer of those measurements doesn't
have to deal with the null case.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
Complementing and extending the changes in #5758.
Add 2 level statistics:
* snapshotCount
* cacheAgeMs
Add 2 counter statistics
* cachedBytes
* WALCompactionTimeMs
snapshotCount can be used to measure transient write errors that are causing snapshots to accumulate
cacheAgeMs can be used to guage the level of write activity into the cache
The differences between cachedBytes stats sampled at different times can be used to calculate cache throughput rates
The ratio (cachedBytes-diskBytes)/WALCompactionTimeMs can be used calculate WAL compaction throughput.
The ratio of difference between first and last WAL compaction time over the interval
length is an estimate of percentage of cache throughput consumed.
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
The description of the cache design was out of date - reflecting an older
design based on checkpoints and evictions. This revision updates the
design to describe snapshots and also clarify that if compaction performance
falls behind the inbound write rate then writes will fail.
Updates based in part of clarifications provided by Jason Wilder. See https://goo.gl/L7AzVu
Signed-off-by: Jon Seymour <jon@wildducktheories.com>
There was a fix in 5b1791, but is not present in the current branch likely due to a rebase issue.
The current code panics with a query like:
select value from cpu group by host order by time desc limit 1
This fixes the panic as well as prevents #5193 from re-occurring. The issue is that agressively
closing the cursors clears out the seeks slice so re-seeking will fail.
Aux iterators now ask the iterator creator what series will be returned
and determine which aux fields to create based on the results.
The `tsdb.Shards` struct also creates a call iterator around the
iterators returned from each shard.
Fill requires an additional function for IteratorCreator to retrieve the
series that will be returned from the iterator. When fill is required
for an aggregate, the IteratorCreator will be asked what series will be
returned by the created iterator.
This reduces some of the lock contention when writing to the cache.
When a new entry is created, it avoids an allocation. It also skips
a check to see if we need to sorted if we already know it needs to sorted.
Writing the snapshot would deduplicate the snapshot points
while still holding the engine write-lock. This can be expensive
under high load and cause writes to back up and OOM the server.
Instead, grab the snapshot under the lock and dedup it after releasing
the lock.
Possible fix for #5442
We were closing the cursor when we read the last block which caused
the internal state to be cleared. In a group by query, we seeked multiple
times so depending on the group by interval and how the data was laid out
in the blocks, we woudl close the cursor and the last block would get skipped.
Fixes#5193
This may be causing slow restart times for systems with many large TSM files.
What I believe is happening at startup in these cases is that multiple goroutines
are started to load each TSM file concurrently. The kernel appears to serialize
mmap calls from the same process so all of the goroutines end up getting blocked
on the actual mmap system call. MAP_POPULATE instruct the kernel to pre-fault the
page table for the files and triggers read-ahead of the pages. For larger, 2GB files,
this makes the mmap call more expensive and slower. When there are many of these files
and calls it is possible to fill all available memory with pagecache. In this case,
the OS will end up pre-faulting pages from one file and have to remove pages that it just
loaded from another files causing slowness. MAP_POPULATE may also be cause much more data
to be pre-faulted than necessary. To load a file, we just need to scan the index at the end
of the file. MAP_POPULATE is likely causing the whole file to be loaded when it won't actually
be accessed for a while (or at all).
Might fix issue #5311.
Some data shapes would cause files to grow larger than the max size more
quickly which resulted in them getting skipped by the full compaction planner
at times. Some datasets that could make this happen are very large keys or
very large numbers of keys (10M). When this happened, multiple max sized
files would accumulate but the blocks would not be full. When the shard went
cold for writes, these files would get recompacted down to the optimal size, but
a lot of space would be wasted in the mean time.
This is contributing to some of the high memory usage on queries and possibly
some OOMs. This is slightly slower, but removing it allows some fairly large
count queries over 5M series to complete instead of crashing the process using
tsm1 engine.
Key() returned the key and the entries. We did not always need the
entries so they would be allocated and ignored. Added a KeyAt func
that just returns the key to avoid the unnecesary entries allocation.
Use sync.Pool for some temporary buffers used while encoding instead of
allocatin new ones each time. Also increased the default buffer size which
might be too small. Probably need to make this a config var.
We were buffering up the data to write into byte slices to reduce
IO calls but at larger sizes, this causes memory to spike. The TSMWriter
was switched to use a bufio.Writer internally so this byte slice buffering
is unnecessary and costly now.
The block count was an uint16 when incrementing the index location
which was an int32. This caused the value the uint16 value to overflow
before the index location was incremented causing the wrong location
to be read on the next iteration of the loop. This triggers the slice
out of range errors.
Added a test that recreates the panic seen in #5257 and possibly #5202 which
is older code.
Fixes#5257
This changes backup and restore to work for TSM. It breaks it for b1 and bz1, but since those are getting removed it's ok.
The backup runs against any host that is specified and can backup either the metasstore, a database, specific retention policy, or a specific shard. It can also take incremental backups with the `since` flag, which will only backup TSM files that have been created since that timestamp.
The backup is safe to run online. However, for shards that are still hot for writes, they won't be able to create new TSM files while the backup for that single shard runs. If the backup isn't too large and the write throughput isn't too high this shouldn't be a problem since the writes will just go into the WAL cache.
This has a few changes in it (unfortuantely). The main change is to run compactions
concurrently. While implementing this, a few query and performance bugs showed up that
are also fixed by this commit.
The test to see if the destination buffer for encoding and decoding a WAL
entry was broken and would cause a panic if there were large batches that
would overflow the buffer size.
Fixes#5075
If the engine is closed while a compaction is going on, the close call
blocks until the goroutine exits. This could be several minutes because
the control does not return back up to the channel selector while there is
still data to write.
* Update compaction to look at newest files of the smallest step first
* Update compaction to look at older files in larger steps if newer files don't have enough small steps to compact
* Changed the TestDefaultCompactionPlanner_CombineSequence test to reflect what's possible now. We'd only have multiple files in the same generation if the all files but one were over the max allowable size.
* Clean up the logic on when full compactions are run and when planning can be skipped
Move the index locations planning to be lazily created after the first
seek when we know what time and direction we're searching for. This
allows files and blocks to be skip before having to scan the files index.
This improves queries times with time filters wherne there are many TSM
files on disk.
* Update cache loader to delete entries from cache
* Add cache.Delete()
* Update delete to look at keys in the Cache in addition to the FileStore
* Update cache compaction to never happen if the cache is empty
MinTime is not in the index for each block so storing it in the block
header is redundant. The encodings also store it in their header so
we are actually storing it 3 times.
Removing this is an incompatible change with the current tsm1 file format.
Added mmap implementation for Windows. It uses MapViewOfFile similar to Bolt's implementation. MapViewOfFile returns a pointer and not a byte array. Bolt changed their data structure to support it.
Instead of changing the implementation of tsm data structure, I used a trick shown in https://groups.google.com/forum/#!topic/golang-nuts/g0nLwQI9www to use SliceHeader to convert the pointer into a slice.
Bolt's implementation also closes the file handle in mmap itself. It was resulting in a timeout, so implemented https://github.com/edsrzf/mmap-go/blob/master/mmap_windows.go logic to keep file handle open until munmap
* Update Plan to do a full compaction if cold for writes
* Remove MaxFileSize as a config variable from Compactor. Should be a set constant
* Update Plan to keep track of if the last check was fully compacted so we can skip future planning calls
* Update compact min file count to 3 so that compactions run more frequently
* remove rolloverTSMFileSize constant that is no longer used
* remove the maxGenerationFileCount since it is no longer a limitation that's necessary with the new compaction scheme. We no longer read WAL segments as part of the compaction so memory is only used as we read in each individual key
* remove minFileCount and switch to a user configurable variable
* remove the mutex from WALSegmentWriter. There's never more than one open in the WAL at one time and it's not exported through any function so the lock on the WAL should be used. This simplified keeping track of the last write time and removed a bunch of unnecessary locks.
* update WALSegmentWriter.Write to take the compressed bytes so that encoding and compression can occur before the call to write (while we don't hold the WAL lock)
* remove a bunch of unnecessary locking in WAL.writeToLog
* Add check for TSM file magic number and vesion
* Remove old tsm, log, and unused cursor code
* Remove references to tsm1dev everywhere except in the inspector
* Clean up config options for compaction and snapshotting
* Remove old TSM configuration options
* Update the config.sample.toml with TSM options
* Update WAL compact to force if it has been cold for writes for a configurable period of time (1h by default)
This was causing races in the code, when the cache was being reloaded,
because back-to-back open-and-closing of the engine during testing left
goroutines running. With this change the engine is completely shutdown
when Close() is called on it.
This change starts by building the sequence of entries, which also
allows the required size of destination buffer to be calculated. Then
the buffer is allocated up-front in 1 call.
Each snapshot and hot value-set is appended to the buffer. If ordering
is violated at anytime, set the 'needSort' flag. Sorting, if necessary,
is performed just before returning the data.
FileStats called frequently during compaction planning was too expensive because
they were cleared out every time a file replaced causing them all to be reloaded.
Insted, we grab the stats that are already maintained by the files themselves from
the files when needed.
First pass at TSM cursor iteration ended up searching the file indexes
too frequently and hurt performance. This changes that to search it once
and then have the cursor hold onto the block locations to seek
to. Doubles the query performance from the first iteration, but still a lot
of room for improvement.
* Add InfluxQLType to Values to map the TSM type to InfluxQL
* Fix bug in WAL where close wouldn't nil out the currentSegment after closing it
* Export writeSnapshot to be used in tests, add argument to run it async or not
* Update reloadCache to load temporary metadata information in the engine
* Update LoadMetadataIndex to use the temp WAL metadata information
Added mmap implementation for Windows. It uses MapViewOfFile similar to Bolt's implementation. MapViewOfFile returns a pointer and not a byte array. Bolt changed their data structure to support it.
Instead of changing the implementation of tsm data structure, I used a trick shown in https://groups.google.com/forum/#!topic/golang-nuts/g0nLwQI9www to use SliceHeader to convert the pointer into a slice.
Something broke with writing to the WAL now that compactions are running
concurrently. There was also a performance problem with Next/Prev doing
twice as many searches as necessary.
Added mmap implementation for Windows based on MapViewOfFile. Used SliceHeader trick to change the pointer returned by MapViewOfFile to a byte slice. This will not call for any change in rest of tsm.
However I am not sure where this mmap function is called, as go build is still complains about
tsdb\engine\tsm1\tsm1.go:1974: undefined: syscall.Mmap
tsdb\engine\tsm1\tsm1.go:1974: undefined: syscall.PROT_READ
tsdb\engine\tsm1\tsm1.go:1974: undefined: syscall.MAP_SHARED
tsdb\engine\tsm1\tsm1.go:2033: undefined: syscall.Munmap
* Update cache to have a single slice of values for a key (removed checkpoints)
* Changed compact.Plan to only worry about TSM files.
* Updated Plan to not return an error since there was no case in which it would.
* Update WAL to not keep stats since they're no longer needed.
* Update engine to flush the Cache/WAL to a new TSM file when the min threshold is hit.
* Split compact logic between TSM compacts and WAL/Cache writes.
* Remove unnecessary merge iterator, wal segment iterator, and other no longer necessary stuff.
* Remove the asending bool from the Dedupe method. Values should always be in ascending order. It's up to the cursor to iterate through values based on the direction. Giving the cursor responsibility makes it so we don't need to sort, dedupe or reallocate anything for different query orders.
* Updated engine to use its locks to ensure writes and cache flushes don't cause a race.
* Update all tests with new signatures. Removed a bunch of tests around TSM rewrites and WAL segment iteration that are no longer necessary.
100mb is easy it hit even with basic stress test config. Don't set
a limit by default so that an operator can size it appropriately based
on their hardware.
This commit fixes a deadlock that occurs during b1 flushes. It's
caused by taking locks in a different order. In the flush, b1
locks the engine and then bolt. However, in the query cursor, a
lock is obtained on bolt first (via `DB.Begin()`) and then the
engine is locked while reading from the engine's cache.
Getting an intermittent test failure with this so removing it for now since compactions
are still able to keep up without it. Will need to look into this further because the
allocations is still very high and will affect compactions over longer periods of time.
MergeIterator will be used to merge multiple TSM KeyIterators and the
WAL KeyIterator using a stream based iteration approach. Each iteration
cycle returns a key and values ordered in way to write a new TSM file
optimally.
This provides and interface and type to combine multiple WAL segments
in order and then allow the values to be read in an order suitable for
writing to a TSM file.
Starting to integrate some of the components into a engine that is
usable for development purposes. This allows the code to evolve while
keeping the existing TSM engine in tact for reference.
Currently, just the WAL is wired up so writes can be tested. Other engine
functions will panic the server if called.
This is the existing WAL + cache implementation. Moving it to a separate file
so that it can remain intact while a refactoring to a independent WAL can occur.
The WAL was also named Log in the code so this names file more closely to the concept
in the code.
This will faciliate loading a block into a type specific result without
first loading the block. This will also allow us to populate the database
index solely from the index.
There is a lot of allocations performed when decoding blocks. These
types can be re-used to reduce allocations in many cases. This change
allows a type specific slice to be passed in to decode funcs to be re-used
if it is large enough.
The existing decode is is left for backwards compatibility but is not
very efficient right now. It may be removed.
Edd Robinson
Jason Wilder
Mark Rushakoff
Jon Seymour
Jason Wilder
Jason Wilder
Jonathan A. Sternberg
Mark Rushakoff
Jonathan A. Sternberg
Joe LeGasse
Ben Johnson
liang@qiniu.com
Justin Nuß
Gabriel Levine
INADA Naoki
runner.mei
Paul Dix
Ady
Alexandre Viau
Philip O'Toole
Philip O'Toole
Fazal Majid
Adarsha