Update sync_data.sh
Browse files- sync_data.sh +26 -28
sync_data.sh
CHANGED
|
@@ -10,6 +10,23 @@ fi
|
|
| 10 |
# 激活虚拟环境
|
| 11 |
. /opt/venv/bin/activate
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
# 上传备份
|
| 14 |
upload_backup() {
|
| 15 |
file_path="$1"
|
|
@@ -17,8 +34,6 @@ upload_backup() {
|
|
| 17 |
|
| 18 |
python3 -c "
|
| 19 |
from huggingface_hub import HfApi
|
| 20 |
-
import sys
|
| 21 |
-
import os
|
| 22 |
api = HfApi(token='$HF_TOKEN')
|
| 23 |
try:
|
| 24 |
api.upload_file(
|
|
@@ -31,40 +46,32 @@ try:
|
|
| 31 |
except Exception as e:
|
| 32 |
print(f'Error uploading file: {str(e)}')
|
| 33 |
"
|
|
|
|
|
|
|
| 34 |
}
|
| 35 |
|
| 36 |
# 下载最新备份
|
| 37 |
download_latest_backup() {
|
| 38 |
python3 -c "
|
| 39 |
from huggingface_hub import HfApi
|
| 40 |
-
import sys
|
| 41 |
-
import os
|
| 42 |
-
import tarfile
|
| 43 |
-
import tempfile
|
| 44 |
api = HfApi(token='$HF_TOKEN')
|
| 45 |
try:
|
| 46 |
files = api.list_repo_files(repo_id='$DATASET_ID', repo_type='dataset')
|
| 47 |
backup_files = [f for f in files if f.startswith('electerm_backup_') and f.endswith('.tar.gz')]
|
| 48 |
-
|
| 49 |
if not backup_files:
|
| 50 |
print('No backup files found')
|
| 51 |
sys.exit()
|
| 52 |
-
|
| 53 |
latest_backup = sorted(backup_files)[-1]
|
| 54 |
-
|
| 55 |
with tempfile.TemporaryDirectory() as temp_dir:
|
| 56 |
filepath = api.hf_hub_download(
|
| 57 |
-
repo_id='$DATASET_ID',
|
| 58 |
-
|
| 59 |
-
repo_type='dataset',
|
| 60 |
-
local_dir=temp_dir
|
| 61 |
)
|
| 62 |
-
|
| 63 |
-
if filepath and os.path.exists(filepath):
|
| 64 |
with tarfile.open(filepath, 'r:gz') as tar:
|
| 65 |
tar.extractall('/app/electerm-web/data')
|
| 66 |
print(f'Successfully restored backup from {latest_backup}')
|
| 67 |
-
|
| 68 |
except Exception as e:
|
| 69 |
print(f'Error downloading backup: {str(e)}')
|
| 70 |
"
|
|
@@ -74,29 +81,20 @@ except Exception as e:
|
|
| 74 |
echo "Downloading latest backup from HuggingFace..."
|
| 75 |
download_latest_backup
|
| 76 |
|
| 77 |
-
#
|
| 78 |
sync_data() {
|
| 79 |
while true; do
|
| 80 |
echo "Starting sync process at $(date)"
|
| 81 |
-
|
| 82 |
if [ -d /app/electerm-web/data ]; then
|
| 83 |
-
|
| 84 |
-
backup_file="electerm_backup_${timestamp}.tar.gz"
|
| 85 |
-
|
| 86 |
-
# 压缩数据目录
|
| 87 |
tar -czf "/tmp/${backup_file}" -C /app/electerm-web/data .
|
| 88 |
-
|
| 89 |
echo "Uploading backup to HuggingFace..."
|
| 90 |
upload_backup "/tmp/${backup_file}" "${backup_file}"
|
| 91 |
-
|
| 92 |
rm -f "/tmp/${backup_file}"
|
| 93 |
else
|
| 94 |
echo "Data directory does not exist yet, waiting for next sync..."
|
| 95 |
fi
|
| 96 |
-
|
| 97 |
-
SYNC_INTERVAL=${SYNC_INTERVAL:-7200}
|
| 98 |
-
echo "Next sync in ${SYNC_INTERVAL} seconds..."
|
| 99 |
-
sleep $SYNC_INTERVAL
|
| 100 |
done
|
| 101 |
}
|
| 102 |
|
|
|
|
| 10 |
# 激活虚拟环境
|
| 11 |
. /opt/venv/bin/activate
|
| 12 |
|
| 13 |
+
# 备份管理函数
|
| 14 |
+
manage_backups() {
|
| 15 |
+
python3 -c "
|
| 16 |
+
from huggingface_hub import HfApi
|
| 17 |
+
api = HfApi(token='$HF_TOKEN')
|
| 18 |
+
try:
|
| 19 |
+
files = api.list_repo_files(repo_id='$DATASET_ID', repo_type='dataset')
|
| 20 |
+
backup_files = sorted([f for f in files if f.startswith('electerm_backup_') and f.endswith('.tar.gz')])
|
| 21 |
+
backup_count = int('$DATASET_N') if '$DATASET_N'.isdigit() else 5
|
| 22 |
+
for file in backup_files[:-backup_count]:
|
| 23 |
+
api.delete_file(path_in_repo=file, repo_id='$DATASET_ID', repo_type='dataset')
|
| 24 |
+
print(f'Deleted old backup: {file}')
|
| 25 |
+
except Exception as e:
|
| 26 |
+
print(f'Error managing backups: {str(e)}')
|
| 27 |
+
"
|
| 28 |
+
}
|
| 29 |
+
|
| 30 |
# 上传备份
|
| 31 |
upload_backup() {
|
| 32 |
file_path="$1"
|
|
|
|
| 34 |
|
| 35 |
python3 -c "
|
| 36 |
from huggingface_hub import HfApi
|
|
|
|
|
|
|
| 37 |
api = HfApi(token='$HF_TOKEN')
|
| 38 |
try:
|
| 39 |
api.upload_file(
|
|
|
|
| 46 |
except Exception as e:
|
| 47 |
print(f'Error uploading file: {str(e)}')
|
| 48 |
"
|
| 49 |
+
|
| 50 |
+
manage_backups
|
| 51 |
}
|
| 52 |
|
| 53 |
# 下载最新备份
|
| 54 |
download_latest_backup() {
|
| 55 |
python3 -c "
|
| 56 |
from huggingface_hub import HfApi
|
| 57 |
+
import tarfile, tempfile, os, sys
|
|
|
|
|
|
|
|
|
|
| 58 |
api = HfApi(token='$HF_TOKEN')
|
| 59 |
try:
|
| 60 |
files = api.list_repo_files(repo_id='$DATASET_ID', repo_type='dataset')
|
| 61 |
backup_files = [f for f in files if f.startswith('electerm_backup_') and f.endswith('.tar.gz')]
|
|
|
|
| 62 |
if not backup_files:
|
| 63 |
print('No backup files found')
|
| 64 |
sys.exit()
|
|
|
|
| 65 |
latest_backup = sorted(backup_files)[-1]
|
|
|
|
| 66 |
with tempfile.TemporaryDirectory() as temp_dir:
|
| 67 |
filepath = api.hf_hub_download(
|
| 68 |
+
repo_id='$DATASET_ID', filename=latest_backup,
|
| 69 |
+
repo_type='dataset', local_dir=temp_dir
|
|
|
|
|
|
|
| 70 |
)
|
| 71 |
+
if os.path.exists(filepath):
|
|
|
|
| 72 |
with tarfile.open(filepath, 'r:gz') as tar:
|
| 73 |
tar.extractall('/app/electerm-web/data')
|
| 74 |
print(f'Successfully restored backup from {latest_backup}')
|
|
|
|
| 75 |
except Exception as e:
|
| 76 |
print(f'Error downloading backup: {str(e)}')
|
| 77 |
"
|
|
|
|
| 81 |
echo "Downloading latest backup from HuggingFace..."
|
| 82 |
download_latest_backup
|
| 83 |
|
| 84 |
+
# 数据同步函数
|
| 85 |
sync_data() {
|
| 86 |
while true; do
|
| 87 |
echo "Starting sync process at $(date)"
|
|
|
|
| 88 |
if [ -d /app/electerm-web/data ]; then
|
| 89 |
+
backup_file="electerm_backup_$(date +%Y%m%d_%H%M%S).tar.gz"
|
|
|
|
|
|
|
|
|
|
| 90 |
tar -czf "/tmp/${backup_file}" -C /app/electerm-web/data .
|
|
|
|
| 91 |
echo "Uploading backup to HuggingFace..."
|
| 92 |
upload_backup "/tmp/${backup_file}" "${backup_file}"
|
|
|
|
| 93 |
rm -f "/tmp/${backup_file}"
|
| 94 |
else
|
| 95 |
echo "Data directory does not exist yet, waiting for next sync..."
|
| 96 |
fi
|
| 97 |
+
sleep ${SYNC_INTERVAL:-7200}
|
|
|
|
|
|
|
|
|
|
| 98 |
done
|
| 99 |
}
|
| 100 |
|