fix

VERSION

@@ -1 +1 @@
-2.1.1,archive_logs interface change
+2.2.2,docker prepare sim - eqt - fix

build/run_build.sh

@@ -10,7 +10,7 @@ if [ "" == "${prj}" ] ; then
     usage
 fi
 
-Cmd="pushd /home/oleg/develop/cvtt2"
+Cmd="pushd /home/oleg/develop/cvtt2-ops"
 Cmd="${Cmd} && (cd ${prj}"
 Cmd="${Cmd} && git pushall)"
 Cmd="${Cmd} && ./build_release.sh -p ${prj}"

docker/market_data/_DEPRECATED_/recorder/md_recorder.sh

@@ -0,0 +1,66 @@
+#!/bin/bash
+
+ValidJobs=('BNBFUT_CLD21' 'BNBSPOT_CLD21' 'COINBASE_CLD21' 'BNBFUT_CVTTDATA' 'BNBSPOT_CVTTDATA' 'COINBASE_CVTTDATA')
+# runs on host to start container
+usage() {
+  echo "Usage: $0 <job_name; one of (${ValidJobs[@]})> [image_tag]"
+  exit 1
+}
+
+is_valid() {
+  local job=$1
+  for valid_job in "${ValidJobs[@]}";
+  do
+    # echo "job=$job valid_job=$valid_job"
+    if [[ "${job}" == "${valid_job}" ]]; then
+      return 0
+    fi
+  done
+  return 1
+}
+
+is_container_running() {
+    local container_name=$1
+
+    if [ "$(docker ps --filter "name=^/${container_name}$" --filter "status=running" -q)" ]; then
+        return 0  # true
+    else
+        return 1  # false
+    fi
+}
+
+
+job=${1}
+if  ! is_valid "${job}"; then
+    usage
+fi
+
+ImageTag=${2}
+if [ "${ImageTag}" == "" ] ; then
+    ImageTag="1.6.9"
+fi
+
+DockerImage=cloud21.cvtt.vpn:5500/md_recorder:${ImageTag}
+ContainerName="md_recorder.${job}"
+
+if is_container_running "$ContainerName"; then
+    echo "Container ${ContainerName} is already running. Aborted."
+    exit
+fi
+
+
+Cmd="docker run"
+Cmd+=" -d"
+Cmd+=" --rm"
+Cmd+=" --network=host"
+# Cmd+=" --pull=always"
+Cmd+=" --name=${ContainerName}"
+Cmd+=" -v /home/cvtt/.creds:/.creds"
+Cmd+=" -v /home/cvtt/prod/data:/app/data"
+Cmd+=" -v /home/cvtt/prod/logs:/logs"
+Cmd+=" -e CONFIG_SERVICE=cloud16.cvtt.vpn:6789"
+Cmd+=" ${DockerImage}"
+Cmd+=" ${job}"
+
+echo ${Cmd}
+eval ${Cmd}

docker/market_data/_DEPRECATED_/recorder/md_recorder_monitor.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+
+usage() {
+  echo "Usage: $0 <instance (CLD21,CVTTDATA)> [<admin_port (def. 7225)>]"
+  exit 1
+}
+
+ValidInstances=('CLD21' 'CVTTDATA')
+is_valid() {
+  local inst=$1
+  for valid_inst in "${ValidInstances[@]}";
+  do
+    if [[ "$inst" == "$valid_inst" ]]; then
+      return 0
+    fi
+  done
+  return 1
+}
+
+is_container_running() {
+    local container_name=$1
+
+    if [ "$(docker ps --filter "name=^/${container_name}$" --filter "status=running" -q)" ]; then
+        return 0  # true
+    else
+        return 1  # false
+    fi
+}
+
+Instance=${1}
+if ! is_valid "${Instance}"; then
+    usage
+fi
+
+AdminPort=7225
+if [ "${2}" != "" ]; then
+    AdminPort=${2}
+fi
+
+ContainerName="md_recorder_monitor.${Instance}"
+
+if is_container_running "$ContainerName"; then
+    echo "Container ${ContainerName} is already running. Aborted."
+    exit
+fi
+
+Cmd="docker run"
+Cmd+=" -d"
+Cmd+=" --rm"
+Cmd+=" --network=host"
+Cmd+=" --pull=always"
+Cmd+=" --name=${ContainerName}"
+Cmd+=" -v /home/cvtt/.creds:/.creds"
+Cmd+=" -v /home/cvtt/prod/logs:/logs"
+Cmd+=" -e CONFIG_SERVICE=cloud16.cvtt.vpn:6789"
+Cmd+=" cloud21.cvtt.vpn:5500/md_recorder_monitor:latest"
+Cmd+=" ${Instance} ${AdminPort}"
+
+echo ${Cmd}
+eval ${Cmd}

research/aggregate_features.sh

@@ -0,0 +1,275 @@
+#!/bin/bash
+
+if [ $# -ne 2 ]; then
+    echo "Usage: $0 <source_database_file> <features_database_file>"
+    exit 1
+fi
+
+SRC_DB=$1
+DEST_DB=$2
+
+if [ ! -f "$SRC_DB" ]; then
+    echo "Error: Source database file $SRC_DB does not exist"
+    exit 1
+fi
+
+echo "Creating feature tables in $DEST_DB using data from $SRC_DB..."
+
+# Create md_1min_trade_features table
+echo "Creating md_1min_trade_features table..."
+sqlite3 "$DEST_DB" "
+DROP TABLE IF EXISTS md_1min_trade_features;
+CREATE TABLE IF NOT EXISTS md_1min_trade_features (
+    bin_tstamp TEXT,
+    tstamp_ns INTEGER,
+    exchange_id TEXT,
+    instrument_id TEXT,
+    price_mean REAL,
+    price_median REAL,
+    volume REAL,
+    vwap REAL,
+    signed_volume REAL,
+    order_flow_imbalance REAL,
+    num_trades INTEGER,
+    avg_trade_size REAL,
+    PRIMARY KEY (bin_tstamp, exchange_id, instrument_id)
+);"
+
+# Create index for md_1min_trade_features
+echo "Creating index for md_1min_trade_features..."
+sqlite3 "$DEST_DB" "
+CREATE UNIQUE INDEX IF NOT EXISTS md_1min_trade_features_uidx
+ON md_1min_trade_features(bin_tstamp, exchange_id, instrument_id);"
+
+# Populate md_1min_trade_features using source database
+echo "Populating md_1min_trade_features..."
+sqlite3 "$SRC_DB" "ATTACH DATABASE '$DEST_DB' AS dest;
+WITH trade_metrics AS (
+    SELECT
+        tstamp,
+        strftime('%Y-%m-%d %H:%M:00', tstamp) as bin_tstamp,
+        exchange_id,
+        instrument_id,
+        px as price,
+        qty,
+        CASE
+            WHEN condition = 'B' THEN qty
+            WHEN condition = 'S' THEN -qty
+            ELSE 0
+        END as signed_qty
+    FROM md_trades
+),
+trade_metrics_agg AS (
+    SELECT
+        bin_tstamp,
+        exchange_id,
+        instrument_id,
+        COUNT(*) as cnt,
+        MIN(tstamp) as min_tstamp,
+        MAX(tstamp) as max_tstamp
+    FROM trade_metrics
+    GROUP BY bin_tstamp, exchange_id, instrument_id
+)
+INSERT INTO dest.md_1min_trade_features
+SELECT
+    tm.bin_tstamp,
+    CAST(strftime('%s', tm.bin_tstamp) * 1000000000 AS INTEGER) as tstamp_ns,
+    tm.exchange_id,
+    tm.instrument_id,
+    AVG(price) as price_mean,
+    AVG(CASE WHEN rank_num >= FLOOR(cnt/2.0) AND rank_num <= CEIL(cnt/2.0) THEN price ELSE NULL END) as price_median,
+    SUM(qty) as volume,
+    SUM(price * qty) / SUM(qty) as vwap,
+    SUM(signed_qty) as signed_volume,
+    SUM(CASE WHEN signed_qty > 0 THEN signed_qty ELSE 0 END) -
+    SUM(CASE WHEN signed_qty < 0 THEN ABS(signed_qty) ELSE 0 END) as order_flow_imbalance,
+    COUNT(*) as num_trades,
+    AVG(qty) as avg_trade_size
+FROM (
+    SELECT
+        tm.*,
+        tma.cnt,
+        ROW_NUMBER() OVER (PARTITION BY tm.bin_tstamp, tm.exchange_id, tm.instrument_id ORDER BY price) as rank_num
+    FROM trade_metrics tm
+    JOIN trade_metrics_agg tma
+        ON tm.bin_tstamp = tma.bin_tstamp
+        AND tm.exchange_id = tma.exchange_id
+        AND tm.instrument_id = tma.instrument_id
+) tm
+GROUP BY tm.bin_tstamp, tm.exchange_id, tm.instrument_id;"
+
+# Create md_1min_quote_features table in destination database
+echo "Creating md_1min_quote_features table..."
+sqlite3 "$DEST_DB" "
+DROP TABLE IF EXISTS md_1min_quote_features;
+CREATE TABLE IF NOT EXISTS md_1min_quote_features (
+    bin_tstamp TEXT,
+    tstamp_ns INTEGER,
+    exchange_id TEXT,
+    instrument_id TEXT,
+    mid_price_open REAL,
+    mid_price_high REAL,
+    mid_price_low REAL,
+    mid_price_close REAL,
+    mid_price_mean REAL,
+    rel_spread_mean REAL,
+    rel_spread_min REAL,
+    rel_spread_max REAL,
+    rel_spread_first REAL,
+    rel_spread_last REAL,
+    l1_imbalance_mean REAL,
+    l1_imbalance_min REAL,
+    l1_imbalance_max REAL,
+    l1_imbalance_first REAL,
+    l1_imbalance_last REAL,
+    micro_price_mean REAL,
+    micro_price_min REAL,
+    micro_price_max REAL,
+    micro_price_first REAL,
+    micro_price_last REAL,
+    weighted_mid_mean REAL,
+    weighted_mid_min REAL,
+    weighted_mid_max REAL,
+    weighted_mid_first REAL,
+    weighted_mid_last REAL,
+    PRIMARY KEY (bin_tstamp, exchange_id, instrument_id)
+);"
+
+# Create index for md_1min_quote_features
+echo "Creating index for md_1min_quote_features..."
+sqlite3 "$DEST_DB" "
+CREATE UNIQUE INDEX IF NOT EXISTS md_1min_quote_features_uidx
+ON md_1min_quote_features(bin_tstamp, exchange_id, instrument_id);"
+
+# Populate md_1min_quote_features using source database
+echo "Populating md_1min_quote_features..."
+sqlite3 "$SRC_DB" "ATTACH DATABASE '$DEST_DB' AS dest;
+INSERT INTO dest.md_1min_quote_features
+SELECT
+    strftime('%Y-%m-%d %H:%M:00', tstamp) as bin_tstamp,
+    CAST(strftime('%s', tstamp) * 1000000000 AS INTEGER) as tstamp_ns,
+    exchange_id,
+    instrument_id,
+    FIRST_VALUE((ask_px + bid_px) / 2.0) OVER w as mid_price_open,
+    MAX((ask_px + bid_px) / 2.0) as mid_price_high,
+    MIN((ask_px + bid_px) / 2.0) as mid_price_low,
+    LAST_VALUE((ask_px + bid_px) / 2.0) OVER w as mid_price_close,
+    AVG((ask_px + bid_px) / 2.0) as mid_price_mean,
+    AVG((ask_px - bid_px) / ((ask_px + bid_px) / 2.0)) as rel_spread_mean,
+    MIN((ask_px - bid_px) / ((ask_px + bid_px) / 2.0)) as rel_spread_min,
+    MAX((ask_px - bid_px) / ((ask_px + bid_px) / 2.0)) as rel_spread_max,
+    FIRST_VALUE((ask_px - bid_px) / ((ask_px + bid_px) / 2.0)) OVER w as rel_spread_first,
+    LAST_VALUE((ask_px - bid_px) / ((ask_px + bid_px) / 2.0)) OVER w as rel_spread_last,
+    AVG((bid_qty - ask_qty) / (bid_qty + ask_qty)) as l1_imbalance_mean,
+    MIN((bid_qty - ask_qty) / (bid_qty + ask_qty)) as l1_imbalance_min,
+    MAX((bid_qty - ask_qty) / (bid_qty + ask_qty)) as l1_imbalance_max,
+    FIRST_VALUE((bid_qty - ask_qty) / (bid_qty + ask_qty)) OVER w as l1_imbalance_first,
+    LAST_VALUE((bid_qty - ask_qty) / (bid_qty + ask_qty)) OVER w as l1_imbalance_last,
+    AVG((ask_px * bid_qty + bid_px * ask_qty) / (bid_qty + ask_qty)) as micro_price_mean,
+    MIN((ask_px * bid_qty + bid_px * ask_qty) / (bid_qty + ask_qty)) as micro_price_min,
+    MAX((ask_px * bid_qty + bid_px * ask_qty) / (bid_qty + ask_qty)) as micro_price_max,
+    FIRST_VALUE((ask_px * bid_qty + bid_px * ask_qty) / (bid_qty + ask_qty)) OVER w as micro_price_first,
+    LAST_VALUE((ask_px * bid_qty + bid_px * ask_qty) / (bid_qty + ask_qty)) OVER w as micro_price_last,
+    AVG((ask_px * ask_qty + bid_px * bid_qty) / (bid_qty + ask_qty)) as weighted_mid_mean,
+    MIN((ask_px * ask_qty + bid_px * bid_qty) / (bid_qty + ask_qty)) as weighted_mid_min,
+    MAX((ask_px * ask_qty + bid_px * bid_qty) / (bid_qty + ask_qty)) as weighted_mid_max,
+    FIRST_VALUE((ask_px * ask_qty + bid_px * bid_qty) / (bid_qty + ask_qty)) OVER w as weighted_mid_first,
+    LAST_VALUE((ask_px * ask_qty + bid_px * bid_qty) / (bid_qty + ask_qty)) OVER w as weighted_mid_last
+FROM md_quotes
+GROUP BY strftime('%Y-%m-%d %H:%M:00', tstamp), exchange_id, instrument_id
+WINDOW w AS (
+    PARTITION BY strftime('%Y-%m-%d %H:%M:00', tstamp), exchange_id, instrument_id
+    ORDER BY tstamp
+    RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING
+);"
+
+# Copy the md_1min_bars table to destination database
+echo "Copying md_1min_bars to destination..."
+sqlite3 "$DEST_DB" "
+DROP TABLE IF EXISTS md_1min_bars;
+CREATE TABLE IF NOT EXISTS md_1min_bars (
+    bin_tstamp TEXT,
+    tstamp_ns INTEGER,
+    exchange_id TEXT,
+    instrument_id TEXT,
+    open REAL,
+    high REAL,
+    low REAL,
+    close REAL,
+    volume REAL,
+    vwap REAL,
+    num_trades INTEGER,
+    PRIMARY KEY (bin_tstamp, exchange_id, instrument_id)
+);"
+
+echo "Creating index for md_1min_bars..."
+sqlite3 "$DEST_DB" "
+CREATE UNIQUE INDEX IF NOT EXISTS md_1min_bars_uidx
+ON md_1min_bars(bin_tstamp, exchange_id, instrument_id);"
+
+echo "Populating md_1min_bars..."
+sqlite3 "$SRC_DB" "ATTACH DATABASE '$DEST_DB' AS dest;
+INSERT INTO dest.md_1min_bars
+SELECT * FROM md_1min_bars;"
+
+# Create the combined features view in destination database
+echo "Creating combined features view..."
+sqlite3 "$DEST_DB" "
+DROP VIEW IF EXISTS md_1min_features_view;
+CREATE VIEW IF NOT EXISTS md_1min_features_view AS
+SELECT
+    b.bin_tstamp,
+    b.tstamp_ns,
+    b.exchange_id,
+    b.instrument_id,
+    -- OHLCV data from md_1min_bars
+    b.open,
+    b.high,
+    b.low,
+    b.close,
+    b.volume,
+    b.vwap,
+    b.num_trades,
+    -- Quote features
+    q.mid_price_open,
+    q.mid_price_high,
+    q.mid_price_low,
+    q.mid_price_close,
+    q.mid_price_mean,
+    q.rel_spread_mean,
+    q.rel_spread_min,
+    q.rel_spread_max,
+    q.rel_spread_first,
+    q.rel_spread_last,
+    q.l1_imbalance_mean,
+    q.l1_imbalance_min,
+    q.l1_imbalance_max,
+    q.l1_imbalance_first,
+    q.l1_imbalance_last,
+    q.micro_price_mean,
+    q.micro_price_min,
+    q.micro_price_max,
+    q.micro_price_first,
+    q.micro_price_last,
+    q.weighted_mid_mean,
+    q.weighted_mid_min,
+    q.weighted_mid_max,
+    q.weighted_mid_first,
+    q.weighted_mid_last,
+    -- Trade features
+    t.price_mean as trade_price_mean,
+    t.price_median as trade_price_median,
+    t.signed_volume,
+    t.order_flow_imbalance,
+    t.avg_trade_size
+FROM md_1min_bars b
+LEFT JOIN md_1min_quote_features q
+    ON b.bin_tstamp = q.bin_tstamp
+    AND b.exchange_id = q.exchange_id
+    AND b.instrument_id = q.instrument_id
+LEFT JOIN md_1min_trade_features t
+    ON b.bin_tstamp = t.bin_tstamp
+    AND b.exchange_id = t.exchange_id
+    AND b.instrument_id = t.instrument_id;"
+
+echo "Feature tables created and populated successfully in $DEST_DB!"

research/load_crypto_md.sh

@@ -3,7 +3,7 @@
 usage() {
   echo -n "Usage: $0 [-h <host (hs01*/cloud21)>]"
   echo -n " [-d <YYYYMMDD> (yesterday*)]"
-  echo -n " [-s <source> (cvttdata/cloud21*)>]"
+  echo -n " [-s <source> (cloud28/cloud21*)>]"
   echo " [-t <target_dir> (/opt/jupyter_gpu/data/crypto_md)]"
   exit 1
 }
@@ -67,7 +67,7 @@ if ! is_valid "${host}"  "${valid_hosts[@]}" ; then
     usage
 fi
 
-valid_sources=('cloud21' 'cvttdata')
+valid_sources=('cloud21' 'cloud28')
 if ! is_valid ${source} ${valid_sources[@]} ; then
     echo "Source '${source}' is not valid"
     usage

scripts/__DEPRECATED__/archive_yesterday_md.sh

@@ -1,44 +0,0 @@
-#!/bin/bash
-
-Python=/home/cvtt/.pyenv/python3.12-venv/bin/python3.12
-RootDir=/home/cvtt/prod
-export PYTHONPATH=${RootDir}
-
-host=${1}
-if [ "${host}" == "cvttdata" ]
-then
-    ArchiveRootDir=/home/cvtt/prod/archive/md_archive/crypto/cvttdata
-    CredKey=TSDB_MD_CVTTDATA_RO
-elif [ "${host}" == "cloud21" ]
-then
-    ArchiveRootDir=/home/cvtt/prod/archive/md_archive/crypto/cloud21
-    CredKey=TSDB_MD_CLD21_RO
-else
-    echo "Unknown host ${host}. ${0} Aborted."
-    exit 1
-fi
-
-mkdir -p ${ArchiveRootDir}
-
-yesterday=$(date -d "yesterday" +%Y%m%d)
-Schemas=${2}
-if [ "${Schemas}" == "" ]
-then
-    Schemas="coinbase,bnbspot,bnbfut"
-fi
-echo "Schemas=${Schemas}"
-
-Cmd=
-Cmd="${Python}"
-Cmd="${Cmd} ${RootDir}/cvttpy/research/utils/archive_ts_md.py"
-Cmd="${Cmd} --config=http://cloud16.cvtt.vpn:6789/apps/md_recorder"
-Cmd="${Cmd} --db_credentials_key=${CredKey}"
-Cmd="${Cmd} --date=${yesterday}"
-Cmd="${Cmd} --schemas=${Schemas}"
-Cmd="${Cmd} --root_dir=${ArchiveRootDir}"
-Cmd="${Cmd} --format=SQLite"
-Cmd="${Cmd} --compress"
-echo ${Cmd}
-eval ${Cmd}
-
-echo "${0} ${*} Done."

scripts/mkt_data/checklists/crypto_md_chcklst.sh

@@ -26,10 +26,10 @@ run_checklist() {
     declare -A Commands
     Commands=(
         ["hs01:cloud21"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cloud21/${yr}/${mn} | tail -5"
-        ["hs01:cvttdata"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cvttdata/${yr}/${mn} | tail -5"
+        ["hs01:cloud28"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cloud28/${yr}/${mn} | tail -5"
         ["hs01:sim"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/sim/ | tail -5"
         ["cloud21:cloud21"]="ssh cvtt@cloud21.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cloud21/${yr}/${mn} | tail -5"
-        ["cloud21:cvttdata"]="ssh cvtt@cloud21.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cvttdata/${yr}/${mn} | tail -5"
+        ["cloud21:cloud28"]="ssh cvtt@cloud21.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cloud28/${yr}/${mn} | tail -5"
         ["cloud21:sim"]="ssh cvtt@cloud21.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/sim | tail -5"
         ["gpushnik"]="ssh oleg@gpushnik.cvtt.vpn 'ls -l /opt/jupyter_gpu/data/crypto_md | tail -10'"
     )

scripts/mkt_data/sim_prepare/prepare_crypto_simdata.sh

@@ -1,112 +1,60 @@
 #!/bin/bash
 
-# SQLite DDL for simulation
-# =========================
-
-
-# -- md_quotes
-
-# -- md_1min_bars
-
-
-usage() {
-  echo -n "Usage: $0"
-  echo -n " -d YYYMMDD Date>"
-  echo -n " [-O <output dir (./) >]"
-  echo -n " [-S <archive_source_host (cloud21)>]"
-  echo -n " [-D <orig_db_source> (cloud21)]"
-  echo
-  exit 1
-}
-
 # --------------------- Settings
-SourceHost=cloud21
+# SOURCE_HOST=cloud21.cvtt.vpn
-SourceRootDir=/opt/store/cvtt/md_archive/crypto
+# SOURCE_ROOT_DIR=/opt/store/cvtt/md_archive/crypto
-DbSource=cloud21
+# DB_SOURCE=cloud28
+# OUTPUT_DIR=/tmp
+# DATE=20250516
+# RSYNC_TARGETS="cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/crypto/sim/ cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/crypto/sim/"
 # --------------------- Settings
 
-while getopts ":d:O:S:D:" opt; do
+if [ -z ${DATE}  ] ; then
-  case ${opt} in
+    DATE=$(date -d 'yesterday' +'%Y%m%d')
-    d )
-      Date=$OPTARG
-      ;;
-    O )
-      OutputDir=$OPTARG
-      ;;
-    S )
-      SourceHost=$OPTARG
-      ;;
-    D )
-      DbSource=$OPTARG
-      ;;
-    \? )
-      echo "Invalid option: -$OPTARG" >&2
-      usage
-      ;;
-    : )
-      echo "Option -$OPTARG requires an argument." >&2
-      usage
-      ;;
-  esac
-done
-
-if [ -z ${Date}  ] ; then
-    echo "date is not specified"
-    usage
-fi
-if [ -z ${OutputDir}  ] ; then
-    OutputDir=.
 fi
 
-SourceHost+=".cvtt.vpn"
+if [ -z ${OUTPUT_DIR}  ] ; then
-mkdir -p ${OutputDir}
+    OUTPUT_DIR=.
+fi
 
-# --- Binance
+echo "DATE=${DATE} SOURCE_HOST=${SOURCE_HOST}"
-Instruments=( PAIR-ADA-USDT )
-Instruments+=( PAIR-BCH-USDT )
-Instruments+=( PAIR-BTC-USDT )
-Instruments+=( PAIR-DOT-USDT )
-Instruments+=( PAIR-ETH-USDT )
-Instruments+=( PAIR-LTC-USDT )
-Instruments+=( PAIR-SOL-USDT )
-Instruments+=( PAIR-USDC-USDT )
-Instruments+=( PAIR-XRP-USDT )
 
-# --- Coinbase
+mkdir -p ${OUTPUT_DIR}
-Instruments+=( PAIR-ADA-USD )
-Instruments+=( PAIR-BCH-USD )
-Instruments+=( PAIR-BTC-USD )
-Instruments+=( PAIR-DOT-USD )
-Instruments+=( PAIR-ETH-USD )
-Instruments+=( PAIR-LTC-USD )
-Instruments+=( PAIR-SOL-USD )
-Instruments+=( PAIR-XRP-USD )
 
-echo "Date=${Date} Instruments=${Instruments[@]} OutputDir=${OutputDir}"
+year=$(date -d ${DATE} +"%Y")
-echo Getting data from ${DataHost} ...
+month=$(date -d ${DATE} +"%m")
+SourceDir="${SOURCE_ROOT_DIR}/${DB_SOURCE}/${year}/${month}"
+SourceFile="${DATE}.mktdata.db.gz"
+SourceFilePath="${SourceDir}/${SourceFile}"
 
-year=$(date -d ${Date} +"%Y")
+Cmd="/usr/bin/rsync -ahv"
-month=$(date -d ${Date} +"%m")
+Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
-SourceDir="${SourceRootDir}/${DbSource}/${year}/${month}"
+Cmd+=" ${SOURCE_HOST}:${SourceFilePath}"
-SourceFile="${SourceDir}/${Date}.mktdata.db.gz"
+Cmd+=" $OUTPUT_DIR/"
-
-Cmd="rsync -ahv"
-Cmd+=" ${SourceHost}:${SourceFile}"
-Cmd+=" $OutputDir/"
 echo ${Cmd}
 eval ${Cmd}
 
-Cmd="(cd ${OutputDir} && gunzip *.db.gz)"
+if [ ! -f ${OUTPUT_DIR}/${SourceFile} ] ; then
+    echo "File ${OUTPUT_DIR}/${SourceFile} NOT FOUND"
+    exit
+fi
+Cmd="(cd ${OUTPUT_DIR} && gunzip -f *.db.gz)"
 echo ${Cmd}
 eval ${Cmd}
 
-SourceDbFile="${OutputDir}/${Date}.mktdata.db"
+SourceDbFile="${OUTPUT_DIR}/${DATE}.mktdata.db"
-ResultDbFile="${OutputDir}/${Date}.crypto_sim_md.db"
+ResultDbFile="${OUTPUT_DIR}/${DATE}.crypto_sim_md.db"
 
+echo "SourceDbFile=${SourceDbFile}"
 echo "Creating Result Database File ${ResultDbFile}"
+cleanup() {
+    rm ${SourceDbFile}
+}
+trap cleanup EXIT
 
 echo "Creating table md_trades ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_trades (
     tstamp text,
     tstamp_ns integer,
@@ -119,12 +67,14 @@ CREATE TABLE  IF NOT EXISTS md_trades (
     condition text,
     tape text
 );
+
 CREATE UNIQUE INDEX  IF NOT EXISTS md_trades_uidx
 ON md_trades(tstamp_ns, exchange_id, instrument_id);
 EOF
 
 echo "Creating table md_quotes ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_quotes (
     tstamp text,
     tstamp_ns integer,
@@ -143,6 +93,7 @@ EOF
 
 echo "Creating table md_1min_bars ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_1min_bars (
 	tstamp text,
 	tstamp_ns integer,
@@ -162,6 +113,7 @@ EOF
 
 echo "Loading md_trades ..."
 sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_trades SELECT
@@ -199,6 +151,7 @@ EOF
 
 echo "Loading md_quotes ..."
 sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_quotes SELECT
@@ -237,6 +190,7 @@ EOF
 
 echo "Loading md_1min_bars ..."
 sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_1min_bars SELECT
@@ -273,25 +227,16 @@ COMMIT;
 DETACH source_db;
 EOF
 
-Cmd="rm ${SourceDbFile}"
-echo ${Cmd}
-eval ${Cmd}
-
-
 Cmd="gzip ${ResultDbFile}"
 echo ${Cmd}
 eval ${Cmd}
 
-Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/crypto/sim/"
+for tgt in ${RSYNC_TARGETS} ; do 
-echo ${Cmd}
+    Cmd="/usr/bin/rsync -ahv"
-eval ${Cmd}
+    Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
-
+    Cmd+=" ${ResultDbFile}.gz ${tgt}"
-Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/crypto/sim/"
+    echo ${Cmd}
-echo ${Cmd}
+    eval ${Cmd}
-eval ${Cmd}
+done 
-
-Cmd="rm ${ResultDbFile}.gz"
-echo ${Cmd}
-eval ${Cmd}
 
 echo Done $0 ${*}

scripts/mkt_data/sim_prepare/prepare_eqt_simdata.sh

@@ -1,13 +1,16 @@
 #!/bin/bash
 
-usage() {
-  echo "Usage: $0  -S <symbols> -d <YYYYMMDD Date> [-O <output dir (./) >]"
-  exit 1
-}
-
 # --------------------- Settings
-SourceHost=cloud21.cvtt.vpn
+SOURCE_HOST=cloud21.cvtt.vpn
-SourceRootDir=/opt/store/cvtt/md_archive/equity/alpaca_md
+SOURCE_ROOT_DIR=/opt/store/cvtt/md_archive/equity/alpaca_md
+SYMBOL_LIST=CAN #,COIN,GBTC,HOOD,MSTR,PYPL,XYZ
+OUTPUT_DIR=/tmp
+DATE=20250514
+RSYNC_TARGETS="cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/equity/test/alpaca_md/sim/"
+# RSYNC_TARGETS="cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/equity/alpaca_md/sim/ 
+#cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/equity/alpaca_md/sim/"
+# --------------------- Settings
+
 # --------------------- Settings
 
 
@@ -27,83 +30,56 @@ is_business_day() {
 }
 export -f is_business_day
 
-
+if [ -z ${SYMBOL_LIST}  ] ; then
-while getopts ":d:S:O:" opt; do
-  case ${opt} in
-    d )
-      Date=$OPTARG
-      ;;
-    S )
-      SymList=$OPTARG
-      ;;
-    O )
-      OutputDir=$OPTARG
-      ;;
-    \? )
-      echo "Invalid option: -$OPTARG" >&2
-      usage
-      ;;
-    : )
-      echo "Option -$OPTARG requires an argument." >&2
-      usage
-      ;;
-  esac
-done
-
-if [ -z ${SymList}  ] ; then
     echo "symbols are not specified"
-    usage
+    exit
 fi
-if [ -z ${Date}  ] ; then
+if [ -z ${DATE}  ] ; then
-    echo "date is not specified"
+    DATE=$(date -d 'yesterday' +'%Y%m%d')
-    usage
 fi
-if [ -z ${OutputDir}  ] ; then
-    OutputDir=.
-fi
-mkdir -p ${OutputDir}
 
-if ! is_business_day ${Date}; then
+mkdir -p ${OUTPUT_DIR}
-   echo "${Date} is not business day"
+
-   usage
+if ! is_business_day ${DATE}; then
+   echo "${DATE} is not business day"
+   exit
 fi
 
 OLD_IFS=${IFS}
 IFS=","
-read -ra Symbols <<< "${SymList}"
+read -ra Symbols <<< "${SYMBOL_LIST}"
 IFS=${OLD_IFS}
 
-echo "Date=${Date} Symbols=${Symbols[@]} OutputDir=${OutputDir}"
+echo "DATE=${DATE} Symbols=${Symbols[@]} OUTPUT_DIR=${OUTPUT_DIR}"
 echo Getting data from ${DataHost} ...
 
-year=$(date -d ${Date} +"%Y")
+year=$(date -d ${DATE} +"%Y")
 for sym in ${Symbols[@]}; do
     inst_id="STOCK-${sym}"
     capital=${sym:0:1}
-    SourceDir="${SourceRootDir}/${year}/${capital}/${sym}"
+    SourceDir="${SOURCE_ROOT_DIR}/${year}/${capital}/${sym}"
-    SourceHbarFile="${SourceDir}/${Date}.${sym}.alpaca_1m_bars.db.gz"
+    SourceHbarFile="${SourceDir}/${DATE}.${sym}.alpaca_1m_bars.db.gz"
-    SourceQatFile="${SourceDir}/${Date}.${sym}.alpaca_qat.db.gz"
+    SourceQatFile="${SourceDir}/${DATE}.${sym}.alpaca_qat.db.gz"
 
     for src_file in ${SourceHbarFile} ${SourceQatFile}; do
         Cmd="rsync -ahv"
-        Cmd+=" ${SourceHost}:${src_file}"
+        Cmd+=" ${SOURCE_HOST}:${src_file}"
-        Cmd+=" $OutputDir/"
+        Cmd+=" $OUTPUT_DIR/"
         echo ${Cmd}
         eval ${Cmd}
     done
 done
 
-Cmd="(cd ${OutputDir} && gunzip *.db.gz)"
+Cmd="(cd ${OUTPUT_DIR} && gunzip *.db.gz)"
 echo ${Cmd}
 eval ${Cmd}
 
-ResultDbFile="${OutputDir}/${Date}.alpaca_sim_md.db"
+ResultDbFile="${OUTPUT_DIR}/${DATE}.alpaca_sim_md.db"
-echo "Creating Result Database File ${ResultDbFile}"
-
 echo "Creating Result Database File ${ResultDbFile}"
 
 echo "Creating table md_trades ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_trades (
     tstamp text,
     tstamp_ns integer,
@@ -122,6 +98,7 @@ EOF
 
 echo "Creating table md_quotes ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_quotes (
     tstamp text,
     tstamp_ns integer,
@@ -161,11 +138,12 @@ EOF
 # set -f # not to expand *
 
 for sym in ${Symbols[@]}; do
-   src_hbar_db=${OutputDir}/${Date}.${sym}.alpaca_1m_bars.db
+   src_hbar_db=${OUTPUT_DIR}/${DATE}.${sym}.alpaca_1m_bars.db
-   src_qat_db=${OutputDir}/${Date}.${sym}.alpaca_qat.db
+   src_qat_db=${OUTPUT_DIR}/${DATE}.${sym}.alpaca_qat.db
 
     echo "Loading md_trades and md_quotes from ${src_qat_db} ..."
     sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${src_qat_db}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_trades SELECT
@@ -200,6 +178,7 @@ EOF
 
     echo "Loading md_1min_bars from ${src_hbar_db} ..."
     sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${src_hbar_db}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_1min_bars SELECT
@@ -220,7 +199,7 @@ COMMIT;
 DETACH source_db;
 EOF
 
-    Cmd="rm ${src_hbar_db} ${src_qat_db}"
+Cmd="rm ${src_hbar_db} ${src_qat_db}"
     echo ${Cmd}
     eval ${Cmd}
 done
@@ -229,16 +208,12 @@ Cmd="gzip ${ResultDbFile}"
 echo ${Cmd}
 eval ${Cmd}
 
-Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/equity/alpaca_md/sim/"
+for tgt in ${RSYNC_TARGETS} ; do 
-echo ${Cmd}
+    Cmd="/usr/bin/rsync -ahv"
-eval ${Cmd}
+    Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
-
+    Cmd+=" ${ResultDbFile}.gz ${tgt}"
-Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/equity/alpaca_md/sim/"
+    echo ${Cmd}
-echo ${Cmd}
+    eval ${Cmd}
-eval ${Cmd}
+done 
-
-Cmd="rm ${ResultDbFile}.gz"
-echo ${Cmd}
-eval ${Cmd}
 
 echo Done $0 ${*}

utils/archive_logs.sh

@@ -11,9 +11,10 @@ function usage {
 
 echo Starting $0 $*
 
-LogDir=${1}
+# ---- D e f a u l t s
-LogArchiveDir=/works/logs
+LogArchiveDir=/works/archive/logs
 DateCriteria="2 days ago"
+# ---- D e f a u l t s
 
 
 # ---------------- cmdline

utils/move_archives.sh

@@ -1,40 +1,57 @@
 #!/bin/bash
-
+# FOR hosts with limited disk space - move to storage server
-# FOR cloud hosts with limited disk space - move to storage server
 
 function usage {
-    echo "Usage: ${0} <host> <from_dir> <days>"
+    echo -n "Usage: ${0}"
+    echo -n " -H <host_label>"
+    echo -n " [ -A <archive_dir> (default /works/archive)]"
+    echo -n " [-D <older than time criteria> (default: '2 days ago')]"
+    echo
     exit 1
 }
 
 echo Starting $0 $*
 
-
+# ---- D e f a u l t s
+ArchiveDir=/works/archive
+DateCriteria="2 days ago"
 FromHost=$(hostname -s)
+# ---- D e f a u l t s
+
+# ---------------- cmdline
+while getopts "A:H:D:" opt; do
+  case ${opt} in
+    A )
+      ArchiveDir=$OPTARG
+      ;;
+    H )
+      FromHost=$OPTARG
+      ;;
+    D )
+      DateCriteria=$OPTARG
+      ;;
+    \? )
+      echo "Invalid option: -$OPTARG" >&2
+      usage
+      ;;
+    : )
+      echo "Option -$OPTARG requires an argument." >&2
+      usage
+      ;;
+  esac
+done
+# ---------------- cmdline
+
+
 if [ "${FromHost}" == "" ]
 then
     usage
 fi
 
-ArchiveDir=${2}
-if [ "${ArchiveDir}" == "" ]
-then
-    usage
-fi
-
-Days=${3}
-if [ "${Days}" == "" ]
-then
-    Days=2
-fi
-
-DateCriteria="${Days} days ago"
-
 TargetHost=cloud21.cvtt.vpn
 TargetRootDir=/opt/store/cvtt/archive
 
-Oldest=$(date -d "${DateCriteria}" '+%Y-%m-%d')
+Oldest=$(date -d "${DateCriteria}" '+%Y-%m-%d %H:%M:%S')
-Now=$(date '+%Y%m%d_%H%M%S')
 
 echo "Looking for log files older than ${DateCriteria} in ${ArchiveDir}"
 Cmd="find ${ArchiveDir}/"
@@ -45,7 +62,7 @@ Cmd="${Cmd} -o -name '*.logs.*'"
 Cmd="${Cmd} -o -name '*.tgz'"
 Cmd="${Cmd} ')'"
 Cmd="${Cmd} -type f"
-Cmd="${Cmd} -not -newermt ${Oldest}"
+Cmd="${Cmd} -not -newermt \"${Oldest}\""
 
 echo ${Cmd}
 files=$(eval ${Cmd})