Browse Source

Further tweak scripts.

master
Roelf Wichertjes 2 years ago
parent
commit
53a7490fca
Signed by: rewby GPG Key ID: 4C2B6D2972EE5423
3 changed files with 41 additions and 17 deletions
  1. +2
    -1
      Dockerfile
  2. +21
    -0
      entrypoint.sh
  3. +18
    -16
      mover.sh

+ 2
- 1
Dockerfile View File

@@ -19,5 +19,6 @@ USER ab
WORKDIR /home/ab

COPY mover.sh /home/ab/mover.sh
COPY entrypoint.sh /home/ab/entrypoint.sh

ENTRYPOINT [ "/tini", "--" ]
ENTRYPOINT [ "/tini", "--", "/home/ab/entrypoint.sh" ]

+ 21
- 0
entrypoint.sh View File

@@ -0,0 +1,21 @@
#!/bin/bash
set -exuo pipefail

# https://stackoverflow.com/a/2173421
trap "trap - SIGTERM && kill -- -$$" SIGINT SIGTERM EXIT

export SHARED_WARCS_DIR="${SHARED_WARCS_DIR:-/data/}"

export INCOMING="${INCOMING:-${SHARED_WARCS_DIR}/incoming/}"
export UPLOAD_QUEUE="${UPLOAD_QUEUE:-${SHARED_WARCS_DIR}/upload-queue/}"

mkdir -pv "${INCOMING}"
mkdir -pv "${UPLOAD_QUEUE}"

case "$1" in
"mover")
./mover.sh
;;

esac


+ 18
- 16
mover.sh View File

@@ -2,22 +2,24 @@

set -euo pipefail

INCOMING="${INCOMING:-/data/incoming/}"
UPLOAD_QUEUE="${UPLOAD_QUEUE:-/data/upload-queue/}"
export INCOMING="${INCOMING:-/data/incoming/}"
export UPLOAD_QUEUE="${UPLOAD_QUEUE:-/data/upload-queue/}"

# find every .warc.gz in the rsync directory
find "${INCOMING}" -type f -not -name ".*"\
| while read filename
do
# skip partial uploads
if [[ $filename =~ rsync-tmp ]]
then
continue
fi
while [[ 1 ]] ; do
# find every .warc.gz in the rsync directory
find "${INCOMING}" -type f -not -name ".*"\
| while read filename
do
# skip partial uploads
if [[ $filename =~ rsync-tmp ]]
then
continue
fi
echo "Moving ${filename}"
mkdir -vp "${UPLOAD_QUEUE}"
echo mv -v "${filename}" "${UPLOAD_QUEUE}/"
done

echo "Moving ${filename}"
mkdir -vp "${UPLOAD_QUEUE}"
mv -v "${filename}" "${UPLOAD_QUEUE}/"
echo "Sleeping 30 seconds..."
sleep 30
done



Loading…
Cancel
Save