.gitignore

@@ -1 +0,0 @@
-.history/

VERSION

@@ -1 +1 @@
-2.3.2
+2.0.2,using HA trading-calendar and config service

build/run_build.sh

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

build_ops.sh

@@ -1,183 +0,0 @@
-#!/usr/bin/env bash
-
-# ---------------- Settings
-
-repo=git@cloud21.cvtt.vpn:/works/git/cvtt2/ops.git
-
-dist_root=/home/cvttdist/software/cvtt2
-dist_user=cvttdist
-dist_host="cloud21.cvtt.vpn"
-dist_ssh_port="22"
-
-dist_locations="cloud21.cvtt.vpn:22 hs01.cvtt.vpn:22"
-version_file="VERSION"
-
-prj=ops
-brnch=master
-interactive=N
-
-# ---------------- Settings
-
-# ---------------- cmdline
-
-usage() {
-  echo "Usage: $0 [-b <branch (master)> -i (interactive)"
-  exit 1
-}
-
-while getopts "b:i" opt; do
-  case ${opt} in
-    b )
-      brnch=$OPTARG
-      ;;
-    i )
-      interactive=Y
-      ;;
-    \? )
-      echo "Invalid option: -$OPTARG" >&2
-      usage
-      ;;
-    : )
-      echo "Option -$OPTARG requires an argument." >&2
-      usage
-      ;;
-  esac
-done
-# ---------------- cmdline
-
-confirm() {
-    if [ "${interactive}" == "Y" ]; then
-        echo "--------------------------------"
-        echo -n "Press <Enter> to continue" && read
-    fi
-}
-
-
-if [ "${interactive}" == "Y" ]; then
-    echo -n "Enter project [${prj}]: "
-    read project
-    if [ "${project}" == "" ]
-    then
-        project=${prj}
-    fi
-else
-    project=${prj}
-fi
-
-# repo=${git_repo_arr[${project}]}
-if [ -z ${repo} ]; then
-    echo "ERROR: Project repository for ${project} not found"
-    exit -1
-fi
-echo "Project repo: ${repo}"
-
-if [ "${interactive}" == "Y" ]; then
-    echo -n "Enter branch to build release from [${brnch}]: "
-    read branch
-    if [ "${branch}" == "" ]
-    then
-        branch=${brnch}
-    fi
-else
-    branch=${brnch}
-fi
-
-tmp_dir=$(mktemp -d)
-function cleanup {
-    cd ${HOME}
-    rm -rf ${tmp_dir}
-}
-trap cleanup EXIT
-
-
-prj_dir="${tmp_dir}/${prj}"
-
-cmd_arr=()
-Cmd="git clone ${repo} ${prj_dir}"
-cmd_arr+=("${Cmd}")
-
-Cmd="cd ${prj_dir}"
-cmd_arr+=("${Cmd}")
-
-if [ "${interactive}" == "Y" ]; then
-    echo "------------------------------------"
-    echo "The following commands will execute:"
-    echo "------------------------------------"
-    for cmd in "${cmd_arr[@]}"
-    do
-        echo ${cmd}
-    done
-fi
-
-confirm
-
-for cmd in "${cmd_arr[@]}"
-do
-    echo ${cmd} && eval ${cmd}
-done
-
-Cmd="git checkout ${branch}"
-echo ${Cmd} && eval ${Cmd}
-if [ "${?}" != "0" ]; then
-    echo "ERROR: Branch ${branch} is not found"
-    cd ${HOME} && rm -rf ${tmp_dir}
-    exit -1
-fi
-
-
-release_version=$(cat ${version_file} | awk -F',' '{print $1}')
-whats_new=$(cat ${version_file} | awk -F',' '{print $2}')
-
-
-echo "--------------------------------"
-echo "Version file: ${version_file}"
-echo "Release version: ${release_version}"
-
-confirm
-
-version_tag="v${release_version}"
-version_comment="'${version_tag} ${project} ${branch} $(date +%Y-%m-%d)\n${whats_new}'"
-
-cmd_arr=()
-Cmd="git tag -a ${version_tag} -m ${version_comment}"
-cmd_arr+=("${Cmd}")
-
-Cmd="git push origin --tags"
-cmd_arr+=("${Cmd}")
-
-Cmd="rm -rf .git"
-cmd_arr+=("${Cmd}")
-
-SourceLoc=../${project}
-
-dist_path="${dist_root}/${project}/${release_version}"
-
-for dist_loc in ${dist_locations}; do
-    dhp=(${dist_loc//:/ })
-    dist_host=${dhp[0]}
-    dist_port=${dhp[1]}
-    Cmd="rsync -avzh"
-    Cmd="${Cmd} --rsync-path=\"mkdir -p ${dist_path}"
-    Cmd="${Cmd} && rsync\" -e \"ssh -p ${dist_ssh_port}\""
-    Cmd="${Cmd} $SourceLoc ${dist_user}@${dist_host}:${dist_path}/"
-    cmd_arr+=("${Cmd}")
-done
-
-if [ "${interactive}" == "Y" ]; then
-    echo "------------------------------------"
-    echo "The following commands will execute:"
-    echo "------------------------------------"
-    for cmd in "${cmd_arr[@]}"
-    do
-        echo ${cmd}
-    done
-fi
-
-confirm
-
-for cmd in "${cmd_arr[@]}"
-do
-    pwd && echo ${cmd} && eval ${cmd}
-done
-
-echo "$0 Done ${project} ${release_version}"

docker/market_data/hist/crypto_md_day.sh

@@ -0,0 +1,27 @@
+#!/bin/bash
+
+host=${1}
+date=${2}
+
+DockerRegistry=cloud21.cvtt.vpn:5500
+DockerImage=${DockerRegistry}/crypto_md_day
+
+if [ -z ${date} ] ; then
+    date=$(date -d "yesterday" +%Y%m%d)
+fi
+
+Cmd="docker run"
+Cmd+=" --pull=always"
+Cmd+=" --network=host"
+Cmd+=" --name=crypto_md_day.${host}.${date}"
+Cmd+=" --rm"
+Cmd+=" --volume=${HOME}/.creds:/.creds"
+Cmd+=" -e CONFIG_SERVICE=cloud16.cvtt.vpn:6789"
+Cmd+=" ${DockerImage}"
+Cmd+=" -h ${host}"
+Cmd+=" -d ${date}"
+# Cmd+=" -s coinbase,bnbspot,bnbfut"
+Cmd+=" -s coinbase,bnbspot"
+
+echo $Cmd
+eval $Cmd

docker/market_data/hist/load_alpaca_hbar.sh

@@ -0,0 +1,107 @@
+#!/bin/bash
+
+usage() {
+  echo "Usage: $0  -N <num_symbols> [-L <LogDir>] [-d <YYYYMMDD Date>]"
+  exit 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
+}
+
+export CalendarURL=https://trading-calendar.cvtt.net/api/v1/markets/hours?mic=XNYS
+
+is_business_day() {
+    dt=${1}
+
+    open_time=$(curl -s "${CalendarURL}&start=${dt}&end=${dt}" | jq '.[] | .open_time')
+    if [ -n  "${open_time}" ]; then
+        return 0
+    else
+        return 1
+    fi
+
+}
+export -f is_business_day
+
+get_prev_business_day() {
+    Start=${1}
+    while true; do
+        if is_business_day ${Start}; then
+            break
+        fi
+        echo "${Start} is not business day in US" >&2
+        Start=$(date -d "${Start} - 1 day" "+%Y-%m-%d")
+    done
+    echo ${Start}
+}
+export -f get_prev_business_day
+
+# ----- Settings
+DockerRegistry=cloud21.cvtt.vpn:5500
+DockerImage=${DockerRegistry}/alpaca_hbar #:latest
+ContainerName=alpaca_hbar
+LogDir=/home/cvtt/prod/logs
+# ----- Settings
+
+while getopts ":d:N:L:" opt; do
+  case ${opt} in
+    d )
+      date_to_load=$OPTARG
+      ;;
+    N )
+      NumSymbols=$OPTARG
+      ;;
+    L )
+      LogDir=$OPTARG
+      ;;
+    \? )
+      echo "Invalid option: -$OPTARG" >&2
+      usage
+      ;;
+    : )
+      echo "Option -$OPTARG requires an argument." >&2
+      usage
+      ;;
+  esac
+done
+
+if is_container_running "$ContainerName"; then
+    echo "Container ${ContainerName} is already running."
+    exit 3
+fi
+
+if [ -z "${date_to_load}" ]; then
+    echo "date_to_load is empty"
+    date_to_load=$(get_prev_business_day $(date  -d "yesterday" '+%Y-%m-%d'))
+    echo "Historical Data for ${date_to_load}"
+fi
+
+echo "date_to_load=${date_to_load}"
+
+Cmd="docker run"
+Cmd+=" --pull=always"
+Cmd+=" --network=host"
+Cmd+=" --name=${ContainerName}"
+Cmd+=" --rm"
+Cmd+=" --volume=${LogDir}:/logs"
+Cmd+=" --volume=${HOME}/.creds:/.creds"
+Cmd+=" -e CONFIG_SERVICE=cloud16.cvtt.vpn:6789"
+Cmd+=" ${DockerImage}"
+Cmd+=" -d ${date_to_load}"
+if [ -n "${NumSymbols}" ]; then
+    Cmd+=" -N ${NumSymbols}"
+fi
+
+echo $Cmd
+eval $Cmd
+
+if [ "$?" != "0" ] ; then
+    exit 1 # if killed we do not save last day
+fi

docker/market_data/hist/load_alpaca_qat.sh

@@ -0,0 +1,110 @@
+#!/bin/bash
+
+usage() {
+  echo "Usage: $0  -S <symbols> [-L <LogDir>] [-d <YYYYMMDD Date>]"
+  exit 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
+}
+
+export CalendarURL=https://trading-calendar.cvtt.net/api/v1/markets/hours?mic=XNYS
+
+is_business_day() {
+    dt=${1}
+
+    open_time=$(curl -s "${CalendarURL}&start=${dt}&end=${dt}" | jq '.[] | .open_time')
+    if [ -n  "${open_time}" ]; then
+        return 0
+    else
+        return 1
+    fi
+
+}
+export -f is_business_day
+
+get_prev_business_day() {
+    Start=${1}
+    while true; do
+        if is_business_day ${Start}; then
+            break
+        fi
+        echo "${Start} is not business day in US" >&2
+        Start=$(date -d "${Start} - 1 day" "+%Y-%m-%d")
+    done
+    echo ${Start}
+}
+export -f get_prev_business_day
+
+# ----- Settings
+DockerRegistry=cloud21.cvtt.vpn:5500
+DockerImage=${DockerRegistry}/alpaca_qat #:latest
+ContainerName=alpaca_qat
+LogDir=/home/cvtt/prod/logs
+# ----- Settings
+
+while getopts ":d:S:L:" opt; do
+  case ${opt} in
+    d )
+      date_to_load=$OPTARG
+      ;;
+    S )
+      Symbols=$OPTARG
+      ;;
+    L )
+      LogDir=$OPTARG
+      ;;
+    \? )
+      echo "Invalid option: -$OPTARG" >&2
+      usage
+      ;;
+    : )
+      echo "Option -$OPTARG requires an argument." >&2
+      usage
+      ;;
+  esac
+done
+
+if [ -z ${Symbols}  ] ; then
+    echo "symbols are not specified"
+    usage
+fi
+
+if is_container_running "$ContainerName"; then
+    echo "Container ${ContainerName} is already running."
+    exit 3
+fi
+
+if [ -z "${date_to_load}" ]; then
+    echo "date_to_load is empty"
+    date_to_load=$(get_prev_business_day $(date  -d "yesterday" '+%Y-%m-%d'))
+    echo "Historical Data for ${date_to_load}"
+fi
+
+echo "date_to_load=${date_to_load}"
+
+Cmd="docker run"
+Cmd+=" --pull=always"
+Cmd+=" --network=host"
+Cmd+=" --name=${ContainerName}"
+Cmd+=" --rm"
+Cmd+=" --volume=${LogDir}:/logs"
+Cmd+=" --volume=${HOME}/.creds:/.creds"
+Cmd+=" -e CONFIG_SERVICE=cloud16.cvtt.vpn:6789"
+Cmd+=" ${DockerImage}"
+Cmd+=" -d ${date_to_load}"
+Cmd+=" -S ${Symbols}"
+
+echo $Cmd
+eval $Cmd
+
+if [ "$?" != "0" ] ; then
+    exit 1 # if killed we do not save last day
+fi

research/aggregate_features.sh

@@ -1,275 +0,0 @@
-#!/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

@@ -1,10 +1,7 @@
 #!/usr/bin/env bash
 
 usage() {
-  echo -n "Usage: $0 [-h <host (hs01*/cloud21)>]"
+  echo "Usage: $0 [-h <host (hs01*/cloud21)>] [-d <YYYYMMDD> (yesterday*)] [-s <source> (cvttdata/cloud21*)>]"
-  echo -n " [-d <YYYYMMDD> (yesterday*)]"
-  echo -n " [-s <source> (cloud28/cloud21*)>]"
-  echo " [-t <target_dir> (/opt/jupyter_gpu/data/crypto_md)]"
   exit 1
 }
 
@@ -22,15 +19,11 @@ is_valid() {
   return 1
 }
 
-# ------- D E F A U L T S
 date=""
 host=hs01
 source=cloud21
-TargetDir="/opt/jupyter_gpu/data/crypto_md"
-# ------- D E F A U L T S
 
-
+while getopts ":h:d:s:" opt; do
-while getopts ":h:d:s:t:" opt; do
   case ${opt} in
     d )
       date=$OPTARG
@@ -41,9 +34,6 @@ while getopts ":h:d:s:t:" opt; do
     s )
       source=$OPTARG
       ;;
-    t )
-      TargetDir=$OPTARG
-      ;;
     \? )
       echo "Invalid option: -$OPTARG" >&2
       usage
@@ -67,7 +57,7 @@ if ! is_valid "${host}"  "${valid_hosts[@]}" ; then
     usage
 fi
 
-valid_sources=('cloud21' 'cloud28')
+valid_sources=('cloud21' 'cvttdata')
 if ! is_valid ${source} ${valid_sources[@]} ; then
     echo "Source '${source}' is not valid"
     usage
@@ -92,6 +82,7 @@ SourceFile=$(date -d ${date} "+%Y%m%d.mktdata.db")
 SourceFileZip="${SourceFile}.gz"
 SourceFilePath=$(date -d ${date} "+${SourceRootDir}/%Y/%m/${SourceFileZip}")
 
+TargetDir="/opt/jupyter_gpu/data/crypto_md"
 TargetFile=$(date -d ${date} "+%Y%m%d.mktdata.ohlcv.db")
 TargetFilePath="${TargetDir}/${TargetFile}"
 

research/prepare_crypto_simdata.sh

@@ -1,94 +1,99 @@
 #!/bin/bash
 
+# SQLite DDL for simulation
+# =========================
+
+
+# -- md_quotes
+
+# -- md_1min_bars
+
+
+usage() {
+  echo "Usage: $0  -d YYYMMDD Date> [-O <output dir (./) >]"
+  exit 1
+}
+
 # --------------------- Settings
-# SOURCE_HOST=cloud21.cvtt.vpn
+SourceHost=cloud21.cvtt.vpn
-# SOURCE_ROOT_DIR=/opt/store/cvtt/md_archive/crypto
+SourceRootDir=/opt/store/cvtt/md_archive/crypto
-# DB_SOURCES=cloud28,cloud29
+DbSource=cloud21
-# 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
 
-if [ -z ${DATE}  ] ; then
+while getopts ":d:O:" opt; do
-    DATE=$(date -d 'yesterday' +'%Y%m%d')
+  case ${opt} in
-fi
+    d )
-
+      Date=$OPTARG
-if [ -z ${OUTPUT_DIR}  ] ; then
+      ;;
-    OUTPUT_DIR=.
+    O )
-fi
+      OutputDir=$OPTARG
-
+      ;;
-echo "DATE=${DATE} SOURCE_HOST=${SOURCE_HOST}"
+    \? )
-
+      echo "Invalid option: -$OPTARG" >&2
-mkdir -p ${OUTPUT_DIR}
+      usage
-
+      ;;
-year=$(date -d ${DATE} +"%Y")
+    : )
-month=$(date -d ${DATE} +"%m")
+      echo "Option -$OPTARG requires an argument." >&2
-
+      usage
-if [ -z "${DB_SOURCES}" ]; then
+      ;;
-    echo "DB_SOURCES is empty"
+  esac
-    exit
-fi
-
-IFS=',' read -r -a db_source_hosts <<< "${DB_SOURCES}"
-
-SourceFile="${DATE}.mktdata.db.gz"
-SelectedSourceHost=""
-SelectedSourceFilePath=""
-SelectedSourceSize=0
-
-for db_source_host in "${db_source_hosts[@]}"; do
-    SourceDir="${SOURCE_ROOT_DIR}/${db_source_host}/${year}/${month}"
-    CandidatePath="${SourceDir}/${SourceFile}"
-    remote_stat_cmd="if [ -f '${CandidatePath}' ]; then stat -c %s '${CandidatePath}'; else exit 1; fi"
-    CandidateSize=$(ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null ${SOURCE_HOST} "${remote_stat_cmd}" 2>/dev/null)
-    if [ $? -eq 0 ] && [ -n "${CandidateSize}" ]; then
-        echo "Found ${SOURCE_HOST}:${CandidatePath} (${CandidateSize} bytes)"
-        if [ -z "${SelectedSourceHost}" ] || [ "${CandidateSize}" -gt "${SelectedSourceSize}" ]; then
-            SelectedSourceHost=${db_source_host}
-            SelectedSourceFilePath=${CandidatePath}
-            SelectedSourceSize=${CandidateSize}
-        fi
-    fi
 done
 
-
+if [ -z ${Date}  ] ; then
-if [ -z "${SelectedSourceHost}" ]; then
+    echo "date is not specified"
-    echo "File ${SourceFile} NOT FOUND on any DB_SOURCES host"
+    usage
-    exit
 fi
+if [ -z ${OutputDir}  ] ; then
+    OutputDir=.
+fi
+mkdir -p ${OutputDir}
 
-echo "Using source ${SelectedSourceHost} with ${SelectedSourceFilePath} (${SelectedSourceSize} bytes)"
+# --- Binance
+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 )
 
-Cmd="/usr/bin/rsync -ahv"
+# --- Coinbase
-Cmd+=" --mkpath"
+Instruments+=( PAIR-ADA-USD )
-Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
+Instruments+=( PAIR-BCH-USD )
-Cmd+=" ${SOURCE_HOST}:${SelectedSourceFilePath}"
+Instruments+=( PAIR-BTC-USD )
-Cmd+=" $OUTPUT_DIR/"
+Instruments+=( PAIR-DOT-USD )
+Instruments+=( PAIR-ETH-USD )
+Instruments+=( PAIR-LTC-USD )
+Instruments+=( PAIR-SOL-USD )
+Instruments+=( PAIR-XRP-USD )
+
+echo "Date=${Date} Instruments=${Instruments[@]} OutputDir=${OutputDir}"
+echo Getting data from ${DataHost} ...
+
+year=$(date -d ${Date} +"%Y")
+month=$(date -d ${Date} +"%m")
+SourceDir="${SourceRootDir}/${DbSource}/${year}/${month}"
+SourceFile="${SourceDir}/${Date}.mktdata.db.gz"
+
+Cmd="rsync -ahv"
+Cmd+=" ${SourceHost}:${SourceFile}"
+Cmd+=" $OutputDir/"
 echo ${Cmd}
 eval ${Cmd}
 
-if [ ! -f ${OUTPUT_DIR}/${SourceFile} ] ; then
+Cmd="(cd ${OutputDir} && gunzip *.db.gz)"
-	echo "File ${OUTPUT_DIR}/${SourceFile} NOT FOUND"
-	exit
-fi
-
-Cmd="(cd ${OUTPUT_DIR} && gunzip -f *.db.gz)"
 echo ${Cmd}
 eval ${Cmd}
 
-SourceDbFile="${OUTPUT_DIR}/${DATE}.mktdata.db"
+SourceDbFile="${OutputDir}/${Date}.mktdata.db"
-ResultDbFile="${OUTPUT_DIR}/${DATE}.crypto_sim_md.db"
+ResultDbFile="${OutputDir}/${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,
@@ -101,14 +106,12 @@ 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,
@@ -127,7 +130,6 @@ 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,
@@ -147,7 +149,6 @@ EOF
 
 echo "Loading md_trades ..."
 sqlite3  ${ResultDbFile} <<EOF
-.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_trades SELECT
@@ -185,7 +186,6 @@ EOF
 
 echo "Loading md_quotes ..."
 sqlite3  ${ResultDbFile} <<EOF
-.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_quotes SELECT
@@ -224,7 +224,6 @@ 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
@@ -261,18 +260,25 @@ COMMIT;
 DETACH source_db;
 EOF
 
+Cmd="rm ${SourceDbFile}"
+echo ${Cmd}
+eval ${Cmd}
+
+
 Cmd="gzip ${ResultDbFile}"
 echo ${Cmd}
 eval ${Cmd}
 
-for tgt in ${RSYNC_TARGETS} ; do 
+Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/crypto/sim/"
-    tgt="${tgt}/${year}/${month}/"
+echo ${Cmd}
-    Cmd="/usr/bin/rsync -ahv"
+eval ${Cmd}
-	Cmd+=" --mkpath"
+
-    Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
+Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/crypto/sim/"
-    Cmd+=" ${ResultDbFile}.gz ${tgt}"
+echo ${Cmd}
-    echo ${Cmd}
+eval ${Cmd}
-    eval ${Cmd}
+
-done 
+Cmd="rm ${ResultDbFile}.gz"
+echo ${Cmd}
+eval ${Cmd}
 
 echo Done $0 ${*}

research/prepare_eqt_simdata.sh

@@ -1,27 +1,23 @@
 #!/bin/bash
 
+usage() {
+  echo "Usage: $0  -S <symbols> -d <YYYYMMDD Date> [-O <output dir (./) >]"
+  exit 1
+}
+
 # --------------------- Settings
-# SOURCE_HOST=cloud21.cvtt.vpn
+SourceHost=cloud21.cvtt.vpn
-# SOURCE_ROOT_DIR=/opt/store/cvtt/md_archive/equity/alpaca_md
+SourceRootDir=/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
 
+
 is_business_day() {
     dt=${1}
     date=$(date -d "${dt}" +"%Y-%m-%d")
 
     CalendarURL=https://trading-calendar.cvtt.net/api/v1/markets/hours?mic=XNYS
     URL="${CalendarURL}&start=${date}&end=${date}"
-    echo "URL=${URL}"
-    # curl $URL
     open_time=$(curl -s "${URL}" | jq '.[] | .open_time')
-    echo "open_time=${open_time}"
-
     if [ -n  "${open_time}" ]; then
         return 0
     else
@@ -31,59 +27,83 @@ 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"
-    exit
+    usage
 fi
-if [ -z ${DATE}  ] ; then
+if [ -z ${Date}  ] ; then
-    DATE=$(date -d 'yesterday' +'%Y%m%d')
+    echo "date is not specified"
+    usage
 fi
+if [ -z ${OutputDir}  ] ; then
+    OutputDir=.
+fi
+mkdir -p ${OutputDir}
 
-mkdir -p ${OUTPUT_DIR}
+if ! is_business_day ${Date}; then
-
+   echo "${Date} is not business day"
-if ! is_business_day ${DATE}; then
+   usage
-   echo "${DATE} is not business day"
-   exit
 fi
 
 OLD_IFS=${IFS}
 IFS=","
-read -ra Symbols <<< "${SYMBOL_LIST}"
+read -ra Symbols <<< "${SymList}"
 IFS=${OLD_IFS}
 
-echo "DATE=${DATE} Symbols=${Symbols[@]} OUTPUT_DIR=${OUTPUT_DIR}"
+echo "Date=${Date} Symbols=${Symbols[@]} OutputDir=${OutputDir}"
 echo Getting data from ${DataHost} ...
 
-year=$(date -d ${DATE} +"%Y")
+year=$(date -d ${Date} +"%Y")
-month=$(date -d ${DATE} +"%m")
-
 for sym in ${Symbols[@]}; do
     inst_id="STOCK-${sym}"
     capital=${sym:0:1}
-    SourceDir="${SOURCE_ROOT_DIR}/${year}/${capital}/${sym}"
+    SourceDir="${SourceRootDir}/${year}/${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+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
+        Cmd+=" ${SourceHost}:${src_file}"
-        Cmd+=" ${SOURCE_HOST}:${src_file}"
+        Cmd+=" $OutputDir/"
-        Cmd+=" $OUTPUT_DIR/"
         echo ${Cmd}
         eval ${Cmd}
     done
 done
 
-Cmd="(cd ${OUTPUT_DIR} && gunzip *.db.gz)"
+Cmd="(cd ${OutputDir} && gunzip *.db.gz)"
 echo ${Cmd}
 eval ${Cmd}
 
-ResultDbFile="${OUTPUT_DIR}/${DATE}.alpaca_sim_md.db"
+ResultDbFile="${OutputDir}/${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,
@@ -102,7 +122,6 @@ EOF
 
 echo "Creating table md_quotes ..."
 sqlite3 ${ResultDbFile} <<EOF
-.echo ON
 CREATE TABLE  IF NOT EXISTS md_quotes (
     tstamp text,
     tstamp_ns integer,
@@ -142,12 +161,11 @@ EOF
 # set -f # not to expand *
 
 for sym in ${Symbols[@]}; do
-   src_hbar_db=${OUTPUT_DIR}/${DATE}.${sym}.alpaca_1m_bars.db
+   src_hbar_db=${OutputDir}/${Date}.${sym}.alpaca_1m_bars.db
-   src_qat_db=${OUTPUT_DIR}/${DATE}.${sym}.alpaca_qat.db
+   src_qat_db=${OutputDir}/${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
@@ -182,7 +200,6 @@ 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
@@ -203,7 +220,7 @@ 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
@@ -212,13 +229,16 @@ Cmd="gzip ${ResultDbFile}"
 echo ${Cmd}
 eval ${Cmd}
 
-for tgt in ${RSYNC_TARGETS} ; do 
+Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/equity/alpaca_md/sim/"
-    Cmd="/usr/bin/rsync -ahv"
+echo ${Cmd}
-    Cmd+=" --mkpath"
+eval ${Cmd}
-    Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
+
-    Cmd+=" ${ResultDbFile}.gz ${tgt}/${year}/${month}/"
+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 ${*}

scripts/__DEPRECATED__/archive_yesterday_md.sh

@@ -0,0 +1,44 @@
+#!/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

@@ -26,11 +26,9 @@ 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:cloud28"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cloud28/${yr}/${mn} | tail -5"
+        ["hs01:cvttdata"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/crypto/cvttdata/${yr}/${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 /opt/store/cvtt/md_archive/crypto/cloud21/${yr}/${mn} | 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 /opt/store/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/checklists/eqty_md_chcklst.sh

@@ -29,14 +29,12 @@ run_checklist() {
         Commands["hs01"]+="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/equity/alpaca_md/${yr}/${sym} | tail -3; "
     done
     Commands["hs01"]+="echo"
-    Commands["hs01:sim"]="ssh cvtt@hs01.cvtt.vpn ls -l /works/cvtt/md_archive/equity/alpaca_md/sim | tail -5" 
 
     Commands["cloud21"]=""
     for sym in ${CheckSymbols}; do
-        Commands["cloud21"]+="ssh cvtt@cloud21.cvtt.vpn ls -l /works/cvtt/md_archive/equity/alpaca_md/${yr}/${sym} | tail -3; "
+        Commands["cloud21"]+="ssh cvtt@cloud21.cvtt.vpn ls -l /opt/store/cvtt/md_archive/equity/alpaca_md/${yr}/${sym} | tail -3; "
     done
     Commands["cloud21"]+="echo"
-    Commands["cloud21:sim"]="ssh cvtt@cloud21.cvtt.vpn ls -l /works/cvtt/md_archive/equity/alpaca_md/sim | tail -5" 
 
     Commands["gpushnik"]="ssh oleg@gpushnik.cvtt.vpn 'ls -l /opt/jupyter_gpu/data/eqty_md | tail -10'"
 
@@ -48,13 +46,6 @@ run_checklist() {
     done
 }
 
-tmpfile=$(mktemp)
-function cleanup {
-    cd ${HOME}
-    rm -f ${tmpfile}
-}
-trap cleanup EXIT
-
 # run_checklist
 
 tmpfile=$(mktemp)

utils/archive_logs.sh

@@ -1,63 +1,36 @@
 #!/bin/bash
 
+
 function usage {
-    echo -n "Usage: ${0}"
+    echo "Usage: ${0}  <log directory> [days (default 2)]"
-    echo -n " -L <log directory>"
-    echo -n " [ -A <archive_logs_dir> (default /works/archive/logs)]"
-    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
+LogDir=${1}
-LogArchiveDir=/works/archive/logs
-DateCriteria="2 days ago"
-# ---- D e f a u l t s
-
-
-# ---------------- cmdline
-while getopts "A:L:D:" opt; do
-  case ${opt} in
-    A )
-      LogArchiveDir=$OPTARG
-      ;;
-    L )
-      LogDir=$OPTARG
-      ;;
-    D )
-      DateCriteria=$OPTARG
-      ;;
-    \? )
-      echo "Invalid option: -$OPTARG" >&2
-      usage
-      ;;
-    : )
-      echo "Option -$OPTARG requires an argument." >&2
-      usage
-      ;;
-  esac
-done
-# ---------------- cmdline
-
-
 if [ "${LogDir}" == "" ]
 then
     usage
 fi
 
-Cmd="mkdir -p ${LogArchiveDir}"
+Days=${2}
-echo ${Cmd} && eval ${Cmd}
+if [ "${Days}" == "" ]
+then
+    Days=2
+fi
 
-echo "Looking for log files older than '${DateCriteria}' in ${LogDir}"
+DateCriteria="${Days} days ago"
 
-Oldest=$(date -d "${DateCriteria}" '+%Y-%m-%d %H:%M:%S')
+Oldest=$(date -d "${DateCriteria}" '+%Y-%m-%d')
 
-Cmd="find ${LogDir}/ '(' -name '*.log' -o -name '*.log.*' ')'  -type f -not -newermt \"${Oldest}\""
+echo "Looking for log files older than ${DateCriteria} in ${LogDir}"
-echo $Cmd
 
-files=$(eval ${Cmd})
+LogArchiveDir=${HOME}/prod/archive/logs
+mkdir -p ${LogArchiveDir}
+
+echo "find ${LogDir}/ '(' -name '*.log' -o -name '*.log.*' ')'  -type f -not -newermt ${Oldest})"
+files=$(find ${LogDir}/ '(' -name '*.log' -o -name '*.log.*' ')'  -type f -not -newermt ${Oldest})
 if [ "$files" == "" ]
 then
     echo "No files found older than ${Oldest} in ${LogDir}"

utils/move_archives.sh

@@ -1,57 +1,40 @@
 #!/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 -n "Usage: ${0}"
+    echo "Usage: ${0} <host> <from_dir> <days>"
-    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 %H:%M:%S')
+Oldest=$(date -d "${DateCriteria}" '+%Y-%m-%d')
+Now=$(date '+%Y%m%d_%H%M%S')
 
 echo "Looking for log files older than ${DateCriteria} in ${ArchiveDir}"
 Cmd="find ${ArchiveDir}/"
@@ -62,7 +45,7 @@ Cmd="${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})