Code

Replay Proto-X#32

Code

Merged

17zhangw merged 102 commits intocmu-db:maincmu-db/dbgym:mainfrom wangpatrick57:replay-protoxwangpatrick57/dbgym:replay-protoxCopy head branch name to clipboardMay 30, 2024

+1,186-640

Commits

Commits on Apr 15, 2024

• copied replay_mythril.py over

  wangpatrick57

  committed
  5ea0984

• added replay function

  wangpatrick57

  committed
  1664890

• Baseilne -> Baseline

  wangpatrick57

  committed
  6dce17a

• pqt -> query_timeout

  wangpatrick57

  committed
  59c3d34

Commits on Apr 16, 2024

• repository -> tuning_steps

  wangpatrick57

  committed
  ae99b21

• removed logdir entirely

  wangpatrick57

  committed
  08829d8

• got rid of output_log_dir entirely

  wangpatrick57

  committed
  cd92d82

• ray results now in dbgym workspace

  wangpatrick57

  committed
  207097a

• now linking hpo-ed params in symlinks

  wangpatrick57

  committed
  93b0988

• now linking tuning steps

  wangpatrick57

  committed
  d2bb709

• replay main working

  wangpatrick57

  committed
  7cd260f

• wrote extract_from_task_run_fordpath

  wangpatrick57

  committed
  aa9b98f

Commits on Apr 17, 2024

• now finding all replay dirs

  wangpatrick57

  committed
  caf0d6c

• added all configs to replay

  wangpatrick57

  committed
  64aaf15

• added replayargs and deleted front of replay_step()

  wangpatrick57

  committed
  6ece74d

• now copying params.json directly into data/

  wangpatrick57

  committed
  e341956

• now copying params.json into tuning_steps

  wangpatrick57

  committed
  a531c2b

• merged with integrate-boot

  wangpatrick57

  committed
  b5f1e8e

• renamed boot_config_fpath to hpo_boot_config_fpath

  wangpatrick57

  committed
  4471787

• added hpo config fpath config to tune

  wangpatrick57

  committed
  315a69f

• fixed bugs so that hpo runs

  wangpatrick57

  committed
  4e9bde6

• fixed some comments

  wangpatrick57

  committed
  17dece3

• made it past first output.log loop

  wangpatrick57

  committed
  2f554e2

• now only reading folders in first loop

  wangpatrick57

  committed
  f035e01

• fixed threshold limit

  wangpatrick57

  committed
  19310a3

• can now build PostgresEnv

  wangpatrick57

  committed
  e084cc7

• now resetting and getting min reward

  wangpatrick57

  committed
  0c3c146

Commits on Apr 18, 2024

• single to double quotes

  wangpatrick57

  committed
  bac5238

• maximal fixed

  wangpatrick57

  committed
  79cee72

• num lines

  wangpatrick57

  committed
  86acc80

• initial fix to run_sample()

  wangpatrick57

  committed
  a3038b5

• fixed all parsing errors

  wangpatrick57

  committed
  551fd67

• run raw csv path fixed

  wangpatrick57

  committed
  0e2486b

• maximal_only fixed

  wangpatrick57

  committed
  55faf8e

• now properly ignoring baseline

  wangpatrick57

  committed
  92020c3

• now parsing action.json

  wangpatrick57

  committed
  98d549b

• now reading prior_state.pkl correctly

  wangpatrick57

  committed
  41a4ac1

• now outputting IndexAction instead of SQL string to action.txt

  wangpatrick57

  committed
  132fb16

• done with combining index acts from action and previous

  wangpatrick57

  committed
  2646f73

• done with combining index acts from action and previous

  wangpatrick57

  committed
  d49bdef

• done with creating index_modifaction_sqls

  wangpatrick57

  committed
  0e3777f

• done with shift_state

  wangpatrick57

  committed
  da8d078

• run_sample running

  wangpatrick57

  committed
  41a9059

• removed indexes from constraints

  wangpatrick57

  committed
  1fcca5a

• 0.1 experiments

  wangpatrick57

  committed
  00b0c87

• only stashing results for tune, and setting idx_name based on index contents now

  wangpatrick57

  committed
  5cf5ef6

• added some comments about idx_name

  wangpatrick57

  committed
  ca0bf85

• now always dumping page cache

  wangpatrick57

  committed
  d92707c

• removed print statements

  wangpatrick57

  committed
  82f9ea1

• duration -> trial_duration

  wangpatrick57

  committed
  c2ad745

• added separate CLI arg for tune duration

  wangpatrick57

  committed
  31f0521

• added print statements to investigate replay behavior

  wangpatrick57

  committed
  9dcd36b

Commits on Apr 19, 2024

• timeout -> workload_timeout

  wangpatrick57

  committed
  3175716

• got rid of modifying workload_timeout

  wangpatrick57

  committed
  43ba9c3

• added tuningmode enum

  wangpatrick57

  committed
  019b4fc

• is_hpo -> tuning_mode

  wangpatrick57

  committed
  02049c4

• replaced replay in pg_env with tuning_mode

  wangpatrick57

  committed
  e7012e1

• changed HPO params to use enums instead of having different names

  wangpatrick57

  committed
  243411c

• hpo, tune, and replay all now not crashing

  wangpatrick57

  committed
  cbb87a7

• added workload timeout during replay param

  wangpatrick57

  committed
  d88fd98

• fixed race condition in multiple threads writing to pg.log

  wangpatrick57

  committed
  bc66526

• now linking to params.json for manual run_*/ traversal

  wangpatrick57

  committed
  f09d38f

• renamed reward in replay.py

  wangpatrick57

  committed
  867aa92

• more renaming

  wangpatrick57

  committed
  74d70ff

• comment changes

  wangpatrick57

  committed
  0e3aa51

• got rid of the 2 maximal params, 2 threshold params, and the 'samples' param

  wangpatrick57

  committed
  fbbee89

• got rid of extra row at bottom

  wangpatrick57

  committed
  106d4ea

• has_timeout -> did_any_query_timeout_in_original

  wangpatrick57

  committed
  7fc0bee

• comment

  wangpatrick57

  committed
  b11d3f4

• refactored codebase so that all symlinks end with .link. full benchmark -> dbms -> tune run working

  wangpatrick57

  committed
  84bcd79

Commits on Apr 22, 2024

• now writing all holon action variations to action.pkl

  wangpatrick57

  committed
  8ee373d

Commits on Apr 23, 2024

• now checking equality with the index space

  wangpatrick57

  committed
  d46c5a9

• added comments describing why query timeout and workload timeout aren't reverse engineer-able

  wangpatrick57

  committed
  95be6fa

• now reliably getting did_any_query_time_out_in_original

  wangpatrick57

  committed
  a909f1b

• fixed did_workload_time_out_in_original and ignoring penalty in original_runtime

  wangpatrick57

  committed
  586b9a3

• changes to scripts

  wangpatrick57

  committed
  d91cc65

• merge

  wangpatrick57

  committed
  aa1bca0

• removing breaking after 10 iterations

  wangpatrick57

  committed
  cc11a8d

• workload_time -> workload_runtime_accum

  wangpatrick57

  committed
  c200e46

• workload_timeout -> this_execution_workload_timeout

  wangpatrick57

  committed
  6beea71

• removed time_left since it's redundant with workload_runtime_accum

  wangpatrick57

  committed
  006cc4a

• removed disable_pg_hint code

  wangpatrick57

  committed
  a12348d

• removed noop index dead code

  wangpatrick57

  committed
  7320e06

• removed dead var

  wangpatrick57

  committed
  e1c3f07

• renamed BestQueryRun.timeout to timed_out

  wangpatrick57

  committed
  9c45bf7

• renamed stop_running to workload_timed_out

  wangpatrick57

  committed
  509f7dc

• refactored execute_workload() to separately return whether the workload or the query timed out

  wangpatrick57

  committed
  22617e0

• replaced workload_runtime_accum with compute_total_workload_runtime()

  wangpatrick57

  committed
  bf5fe73

• now seeing whether workload or query timed out in replay

  wangpatrick57

  committed
  6d237ec

• now logging this_step_run_data before validity checks

  wangpatrick57

  committed
  5bd43c6

Commits on Apr 24, 2024

• added replay_all_variations option

  wangpatrick57

  committed
  c6b15dd

• added comments to _mutilate_action_with_metrics

  wangpatrick57

  committed
  d0ed37f

• added comment about best observed in replay.py

  wangpatrick57

  committed
  b64fda2

• changed bool of queries timed out to an actual num

  wangpatrick57

  committed
  4fab4f2

Commits on Apr 25, 2024

• added info for num executed queries

  wangpatrick57

  committed
  a35a576

• reset now doesn't overwrite the results from step

  wangpatrick57

  committed
  6016334

• wrote load_per_machine_envvars.sh

  wangpatrick57

  committed
  4736315

• added build_space_good_for_boot option

  wangpatrick57

  committed
  9849a99

Commits on Apr 28, 2024

• resolved some PR comments

  wangpatrick57

  committed
  d2fb275

Commits on May 27, 2024

• added comment about tune

  wangpatrick57

  committed
  af33bc7

• different tune trials during hpo now name their tuning_steps dir differently

  wangpatrick57

  committed
  474d7ee

Commits on May 30, 2024

• now logging during HPO for both baseline and tuning steps

  wangpatrick57

  committed
  a6e00b9