prometheus

Commit Graph

Author	SHA1	Message	Date
Ganesh Vernekar	d4b9fe801f	M-map full chunks of Head from disk (#6679 ) When appending to the head and a chunk is full it is flushed to the disk and m-mapped (memory mapped) to free up memory Prom startup now happens in these stages - Iterate the m-maped chunks from disk and keep a map of series reference to its slice of mmapped chunks. - Iterate the WAL as usual. Whenever we create a new series, look for it's mmapped chunks in the map created before and add it to that series. If a head chunk is corrupted the currpted one and all chunks after that are deleted and the data after the corruption is recovered from the existing WAL which means that a corruption in m-mapped files results in NO data loss. [Mmaped chunks format](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/head_chunks.md) - main difference is that the chunk for mmaping now also includes series reference because there is no index for mapping series to chunks. [The block chunks](https://github.com/prometheus/prometheus/blob/master/tsdb/docs/format/chunks.md) are accessed from the index which includes the offsets for the chunks in the chunks file - example - chunks of series ID have offsets 200, 500 etc in the chunk files. In case of mmaped chunks, the offsets are stored in memory and accessed from that. During WAL replay, these offsets are restored by iterating all m-mapped chunks as stated above by matching the series id present in the chunk header and offset of that chunk in that file. Prombench results _WAL Replay_ 1h Wal reply time 30% less wal reply time - 4m31 vs 3m36 2h Wal reply time 20% less wal reply time - 8m16 vs 7m _Memory During WAL Replay_ High Churn: 10-15% less RAM - 32gb vs 28gb 20% less RAM after compaction 34gb vs 27gb No Churn: 20-30% less RAM - 23gb vs 18gb 40% less RAM after compaction 32.5gb vs 20gb Screenshots are in [this comment](https://github.com/prometheus/prometheus/pull/6679#issuecomment-621678932) Signed-off-by: Ganesh Vernekar <cs15btech11018@iith.ac.in>	5 years ago
Julien Pivotto	653f343547	Revert head posting optimization This reverts commit `52630ad0c7`. Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>	5 years ago
Bartlomiej Plotka	c4eefd1b3a	storage: Removed SelectSorted method; Simplified interface; Added requirement for remote read to sort response. This is technically BREAKING CHANGE, but it was like this from the beginning: I just notice that we rely in Prometheus on remote read being sorted. This is because we use selected data from remote reads in MergeSeriesSet which rely on sorting. I found during work on https://github.com/prometheus/prometheus/pull/5882 that we do so many repetitions because of this, for not good reason. I think I found a good balance between convenience and readability with just one method. Smaller the interface = better. Also I don't know what TestSelectSorted was testing, but now it's testing sorting. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>	5 years ago
Julien Pivotto	52630ad0c7	Make head Postings only return series in time range benchmark old ns/op new ns/op delta BenchmarkQuerierSelect/Head/1of1000000-8 405805161 120436132 -70.32% BenchmarkQuerierSelect/Head/10of1000000-8 403079620 120624292 -70.07% BenchmarkQuerierSelect/Head/100of1000000-8 404678647 120923522 -70.12% BenchmarkQuerierSelect/Head/1000of1000000-8 403145813 118636563 -70.57% BenchmarkQuerierSelect/Head/10000of1000000-8 405020046 125716206 -68.96% BenchmarkQuerierSelect/Head/100000of1000000-8 426305002 175808499 -58.76% BenchmarkQuerierSelect/Head/1000000of1000000-8 619002108 567013003 -8.40% BenchmarkQuerierSelect/SortedHead/1of1000000-8 1276316086 120281094 -90.58% BenchmarkQuerierSelect/SortedHead/10of1000000-8 1282631170 121836526 -90.50% BenchmarkQuerierSelect/SortedHead/100of1000000-8 1325824787 121174967 -90.86% BenchmarkQuerierSelect/SortedHead/1000of1000000-8 1271386268 121025117 -90.48% BenchmarkQuerierSelect/SortedHead/10000of1000000-8 1280223345 130838948 -89.78% BenchmarkQuerierSelect/SortedHead/100000of1000000-8 1271401620 243635515 -80.84% BenchmarkQuerierSelect/SortedHead/1000000of1000000-8 1360256090 1307744674 -3.86% BenchmarkQuerierSelect/Block/1of1000000-8 748183120 707888498 -5.39% BenchmarkQuerierSelect/Block/10of1000000-8 741084129 716317249 -3.34% BenchmarkQuerierSelect/Block/100of1000000-8 722157273 735624256 +1.86% BenchmarkQuerierSelect/Block/1000of1000000-8 727587744 731981838 +0.60% BenchmarkQuerierSelect/Block/10000of1000000-8 727518578 726860308 -0.09% BenchmarkQuerierSelect/Block/100000of1000000-8 765577046 757382386 -1.07% BenchmarkQuerierSelect/Block/1000000of1000000-8 1126722881 1084779083 -3.72% benchmark old allocs new allocs delta BenchmarkQuerierSelect/Head/1of1000000-8 4000018 24 -100.00% BenchmarkQuerierSelect/Head/10of1000000-8 4000036 82 -100.00% BenchmarkQuerierSelect/Head/100of1000000-8 4000216 625 -99.98% BenchmarkQuerierSelect/Head/1000of1000000-8 4002016 6028 -99.85% BenchmarkQuerierSelect/Head/10000of1000000-8 4020016 60037 -98.51% BenchmarkQuerierSelect/Head/100000of1000000-8 4200016 600047 -85.71% BenchmarkQuerierSelect/Head/1000000of1000000-8 6000016 6000016 +0.00% BenchmarkQuerierSelect/SortedHead/1of1000000-8 4000055 28 -100.00% BenchmarkQuerierSelect/SortedHead/10of1000000-8 4000073 87 -100.00% BenchmarkQuerierSelect/SortedHead/100of1000000-8 4000253 630 -99.98% BenchmarkQuerierSelect/SortedHead/1000of1000000-8 4002053 6036 -99.85% BenchmarkQuerierSelect/SortedHead/10000of1000000-8 4020053 60054 -98.51% BenchmarkQuerierSelect/SortedHead/100000of1000000-8 4200053 600074 -85.71% BenchmarkQuerierSelect/SortedHead/1000000of1000000-8 6000053 6000053 +0.00% BenchmarkQuerierSelect/Block/1of1000000-8 6000021 6000021 +0.00% BenchmarkQuerierSelect/Block/10of1000000-8 6000057 6000057 +0.00% BenchmarkQuerierSelect/Block/100of1000000-8 6000417 6000417 +0.00% BenchmarkQuerierSelect/Block/1000of1000000-8 6004017 6004017 +0.00% BenchmarkQuerierSelect/Block/10000of1000000-8 6040017 6040017 +0.00% BenchmarkQuerierSelect/Block/100000of1000000-8 6400017 6400017 +0.00% BenchmarkQuerierSelect/Block/1000000of1000000-8 10000018 10000018 +0.00% benchmark old bytes new bytes delta BenchmarkQuerierSelect/Head/1of1000000-8 176001177 1392 -100.00% BenchmarkQuerierSelect/Head/10of1000000-8 176002329 4368 -100.00% BenchmarkQuerierSelect/Head/100of1000000-8 176013849 33520 -99.98% BenchmarkQuerierSelect/Head/1000of1000000-8 176129056 321456 -99.82% BenchmarkQuerierSelect/Head/10000of1000000-8 177281049 3427376 -98.07% BenchmarkQuerierSelect/Head/100000of1000000-8 188801049 35055408 -81.43% BenchmarkQuerierSelect/Head/1000000of1000000-8 304001059 304001049 -0.00% BenchmarkQuerierSelect/SortedHead/1of1000000-8 229192188 2488 -100.00% BenchmarkQuerierSelect/SortedHead/10of1000000-8 229193340 5568 -100.00% BenchmarkQuerierSelect/SortedHead/100of1000000-8 229204860 35536 -99.98% BenchmarkQuerierSelect/SortedHead/1000of1000000-8 229320060 345104 -99.85% BenchmarkQuerierSelect/SortedHead/10000of1000000-8 230472060 3894672 -98.31% BenchmarkQuerierSelect/SortedHead/100000of1000000-8 241992060 40511632 -83.26% BenchmarkQuerierSelect/SortedHead/1000000of1000000-8 357192060 357192060 +0.00% BenchmarkQuerierSelect/Block/1of1000000-8 227201516 227201506 -0.00% BenchmarkQuerierSelect/Block/10of1000000-8 227203057 227203041 -0.00% BenchmarkQuerierSelect/Block/100of1000000-8 227217161 227217165 +0.00% BenchmarkQuerierSelect/Block/1000of1000000-8 227358279 227358289 +0.00% BenchmarkQuerierSelect/Block/10000of1000000-8 228769485 228769475 -0.00% BenchmarkQuerierSelect/Block/100000of1000000-8 242881487 242881477 -0.00% BenchmarkQuerierSelect/Block/1000000of1000000-8 384001705 384001705 +0.00% Signed-off-by: Julien Pivotto <roidelapluie@inuits.eu>	5 years ago
Bartlomiej Plotka	34426766d8	Unify Iterator interfaces. All point to storage now. This is part of https://github.com/prometheus/prometheus/pull/5882 that can be done to simplify things. All todos I added will be fixed in follow up PRs. * querier.Querier, querier.Appender, querier.SeriesSet, and querier.Series interfaces merged with storage interface.go. All imports that. * querier.SeriesIterator replaced by chunkenc.Iterator * Added chunkenc.Iterator.Seek method and tests for xor implementation (?) * Since we properly handle SelectParams for Select methods I adjusted min max based on that. This should help in terms of performance for queries with functions like offset. * added Seek to deletedIterator and test. * storage/tsdb was removed as it was only a unnecessary glue with incompatible structs. No logic was changed, only different source of abstractions, so no need for benchmarks. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>	5 years ago
Thor	17d8c49919	made stripe size configurable (#6644 ) Signed-off-by: Thor <thansen@digitalocean.com>	5 years ago
Brian Brazil	61262159c4	Simplify benchmark given the new API Signed-off-by: Brian Brazil <brian.brazil@robustperception.io>	5 years ago
Brian Brazil	d682731efc	Extend BenchmarkQuerierSelect to use multiple blocks. Signed-off-by: Brian Brazil <brian.brazil@robustperception.io>	5 years ago
Brian Brazil	1d1732bc25	Add benchmark for Querier.Select over blocks and head. Signed-off-by: Brian Brazil <brian.brazil@robustperception.io>	5 years ago
Brian Brazil	aff9f7a9e8	Extend PostingsForMatchers benchmark to cover Blocks too. Signed-off-by: Brian Brazil <brian.brazil@robustperception.io>	5 years ago

10 Commits (5bb7f00d00ba2d73488630851b352974511c233a)