From b35b07c9d2c48b0503d4fe2e3b1b29851a65931d Mon Sep 17 00:00:00 2001
From: Slach <bloodjazman@gmail.com>
Date: Mon, 20 Nov 2023 19:33:37 +0400
Subject: [PATCH] improve performance with parallel during checksums.txt

---
 Dockerfile                           | 2 +-
 test/integration/integration_test.go | 2 +-
 test/integration/kopia/download.sh   | 6 ++----
 test/integration/kopia/upload.sh     | 6 ++----
 4 files changed, 6 insertions(+), 10 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index 50273133..1c3cd888 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -101,7 +101,7 @@ FROM ${CLICKHOUSE_IMAGE}:${CLICKHOUSE_VERSION} AS image_full
 ARG TARGETPLATFORM
 MAINTAINER Eugene Klimov <eklimov@altinity.com>
 
-RUN apt-get update && apt-get install -y gpg xxd bsdmainutils && wget -qO- https://kopia.io/signing-key | gpg --dearmor -o /usr/share/keyrings/kopia-keyring.gpg && \
+RUN apt-get update && apt-get install -y gpg xxd bsdmainutils parallel && wget -qO- https://kopia.io/signing-key | gpg --dearmor -o /usr/share/keyrings/kopia-keyring.gpg && \
     echo "deb [signed-by=/usr/share/keyrings/kopia-keyring.gpg] http://packages.kopia.io/apt/ stable main" > /etc/apt/sources.list.d/kopia.list && \
     wget -c "https://github.com/mikefarah/yq/releases/latest/download/yq_linux_$(dpkg --print-architecture)" -O /usr/bin/yq && chmod +x /usr/bin/yq && \
     apt-get update -y && \
diff --git a/test/integration/integration_test.go b/test/integration/integration_test.go
index 217cf46d..2429a0b1 100644
--- a/test/integration/integration_test.go
+++ b/test/integration/integration_test.go
@@ -409,7 +409,7 @@ func init() {
 	// kopia
 	r.NoError(dockerExec("clickhouse-backup", "bash", "-ce", "curl -sfL https://kopia.io/signing-key | gpg --dearmor -o /usr/share/keyrings/kopia-keyring.gpg"))
 	r.NoError(dockerExec("clickhouse-backup", "bash", "-ce", "echo 'deb [signed-by=/usr/share/keyrings/kopia-keyring.gpg] https://packages.kopia.io/apt/ stable main' > /etc/apt/sources.list.d/kopia.list"))
-	installDebIfNotExists(r, "clickhouse-backup", "kopia", "xxd", "bsdmainutils")
+	installDebIfNotExists(r, "clickhouse-backup", "kopia", "xxd", "bsdmainutils", "parallel")
 	// restic
 	r.NoError(dockerExec("clickhouse-backup", "bash", "-xec", "RELEASE_TAG=$(curl -H 'Accept: application/json' -sL https://github.com/restic/restic/releases/latest | jq -c -r -M '.tag_name'); RELEASE=$(echo ${RELEASE_TAG} | sed -e 's/v//'); curl -sfL \"https://github.com/restic/restic/releases/download/${RELEASE_TAG}/restic_${RELEASE}_linux_amd64.bz2\" | bzip2 -d > /bin/restic; chmod +x /bin/restic"))
 }
diff --git a/test/integration/kopia/download.sh b/test/integration/kopia/download.sh
index 5bfedb44..e2f78bcc 100755
--- a/test/integration/kopia/download.sh
+++ b/test/integration/kopia/download.sh
@@ -6,12 +6,10 @@ BACKUP_NAME=$1
 ${CUR_DIR}/list.sh | grep "${BACKUP_NAME}" | while IFS= read -r line; do
   SNAPSHOT_ID=$(echo "${line}" | jq -r -c -M .snapshot_id)
   SNAPSHOT_PATH=$(echo "${line}" | jq -r -c -M .snapshot_path)
-  kopia restore "${SNAPSHOT_ID}" --skip-existing ${SNAPSHOT_PATH}
+  kopia restore --parallel=$(nproc) "${SNAPSHOT_ID}" --skip-existing ${SNAPSHOT_PATH}
   LOCAL_BACKUP_DIR="$(dirname ${SNAPSHOT_PATH})/${BACKUP_NAME}"
   rm -rf "${LOCAL_BACKUP_DIR}"
-  find "${SNAPSHOT_PATH}" -type f -name checksums.txt | while read CHECKSUMS_FILE; do
-    "${CUR_DIR}/checksum_parser.sh" "${CHECKSUMS_FILE}" "download" "${SNAPSHOT_PATH}"
-  done
+  find "${SNAPSHOT_PATH}" -type f -name checksums.txt | parallel -j $(nproc) "${CUR_DIR}/checksum_parser.sh" {} "download" "${SNAPSHOT_PATH}"
   # need separately `rm` cause hash file can contains multiple the same files in different parts
   find ${SNAPSHOT_PATH} -maxdepth 1 -type f -regex '.*/[a-z0-9]\{32\}.*' | while read HASH_FILE; do
     rm "${SNAPSHOT_PATH}/${HASH_FILE}"
diff --git a/test/integration/kopia/upload.sh b/test/integration/kopia/upload.sh
index 1189d86e..e7f6b8d1 100755
--- a/test/integration/kopia/upload.sh
+++ b/test/integration/kopia/upload.sh
@@ -15,14 +15,12 @@ for dir in $(echo "${LOCAL_PATHS}"); do
     UPLOAD_DIR="$(dirname "${dir}")/latest"
     rm -rf "${UPLOAD_DIR}"
     cp -rl "${dir}" "${UPLOAD_DIR}"
-    find "${UPLOAD_DIR}" -type f -name checksums.txt | while read CHECKSUMS_FILE; do
-      "${CUR_DIR}/checksum_parser.sh" "${CHECKSUMS_FILE}" "upload" "${UPLOAD_DIR}"
-    done
+    find "${UPLOAD_DIR}" -type f -name checksums.txt | parallel -j $(nproc) "${CUR_DIR}/checksum_parser.sh" {} "upload" "${UPLOAD_DIR}"
     SNAPSHOT_SOURCES="${UPLOAD_DIR} ${SNAPSHOT_SOURCES}"
   fi
 done
 
-kopia snapshot create $DIFF_FROM_REMOTE_CMD --fail-fast --tags="backup_name:${BACKUP_NAME}"  $SNAPSHOT_SOURCES
+kopia snapshot create $DIFF_FROM_REMOTE_CMD  --parallel=$(nproc) --fail-fast --tags="backup_name:${BACKUP_NAME}"  $SNAPSHOT_SOURCES
 
 for dir in $(echo "${LOCAL_PATHS}"); do
   if [[ -d "${dir}" ]]; then