Merge pull request #1497 from SkynetLabs/adjust-blocklist-airtable-script

do not prune nginx cache on airtable blocklist script

docker/nginx/conf.d/server/server.local

@@ -1,6 +1,9 @@
 include /etc/nginx/conf.d/include/init-optional-variables;
 
 location /skynet/blocklist {
+    client_max_body_size 10m; # increase max body size to account for large lists
+    client_body_buffer_size 10m; # force whole body to memory so we can read it
+
     content_by_lua_block {        
         local httpc = require("resty.http").new()
 

scripts/blocklist-skylink.sh

@@ -1,18 +1,15 @@
 #! /usr/bin/env bash
 
-# This script adds a skylink to the sia blocklist and removes the skylink from
+# This script is for manual skylink blocking. It accepts either a single 
-# nginx cache. The script should be run locally on each skynet webportal
+# skylink or a file containing list of skylinks. The script is intented
-# server. The automatic script that is used to continuously sync an Airtable
+# for manual use and it should be run locally on each skynet webportal server. 
-# sheet list with the blocklist on the web portals is
+# The automatic script that is used to continuously sync an Airtable sheet 
-# /setup-scripts/blocklist-airtable.py 
+# list with the blocklist on the web portals is /setup-scripts/blocklist-airtable.py 
 
 set -e # exit on first error
 
-# Number of skylinks to block within one batch
-BATCH_SIZE=1000
-
 if [ -z "$1" ]; then
-    echo "Please provide either a skylink or file with skylinks separated by new lines" && exit 1
+    echo "Please provide either a skylink or a file with skylinks separated by new lines" && exit 1
 fi
 
 #########################################################
@@ -37,45 +34,18 @@ else
     skylinks=("$1") # just single skylink passed as input argument
 fi
 
-# Block skylinks in batches
+# get local nginx ip adress
-skylinks_len=${#skylinks[@]}
+nginx_ip=$(docker inspect -f '{{range.NetworkSettings.Networks}}{{.IPAddress}}{{end}}' nginx)
-for (( i = 0; i < $skylinks_len; i++ )); do
-    # Add skylink to batch
-    skylink="${skylinks[$i]}"
-    echo ".. ⌁ Adding skylink ${skylink} to batch..."
-    batch_skylinks+=("$skylink")
 
-    # For performance reasons on each iteration we do not block a single
+# iterate over provided skylinks and block them one by one
-    # skylink, but we block skylinks in batches with BATCH_SIZE size mainly
+for skylink in "${skylinks[@]}"; do
-    # because of nginx cache search.
+    printf "Blocking ${skylink} ... "
-    # If (batch len == batch size) or (we have last batch):
+    status_code=$(curl --write-out '%{http_code}' --silent --output /dev/null --data "{\"add\":[\"$skylink\"]}" "http://${nginx_ip}:8000/skynet/blocklist")
-    if (( ${#batch_skylinks[@]} == $BATCH_SIZE || $i == $skylinks_len - 1 )); then
-        echo "--------------------------------------------"
 
-        # Add to Sia blocklist
+    # print blocklist response status code
-        echo "Blocking batch skylinks in skyd..."
+    if [ $status_code = "204" ]; then
-        skylinks_space_separated="$(IFS=' '; echo "${batch_skylinks[*]}")"
+        echo "done"
-        docker exec sia siac skynet blocklist add $skylinks_space_separated
+    else
-
+        echo "error $status_code"
-        # Remove from NGINX cache
-        # NOTE:
-        # If there are changes to how the NGINX cache is being cleared, the same
-        # changes need to be applied to the /setup-scripts/blocklist-airtable.py
-        # script.
-        echo "Removing batch skylinks from Nginx cache..."
-        skylinks_pipe_separated="$(IFS='|'; echo "${batch_skylinks[*]}")"
-        cached_files_command="find /data/nginx/cache/ -type f | xargs -r grep -Els '^Skynet-Skylink: ($skylinks_pipe_separated)'"
-        docker exec -it nginx bash -c "${cached_files_command} | xargs -r rm"
-
-        # Clear batch
-        batch_skylinks=()
-
-        echo "--------------------------------------------"
     fi
 done
-
-# Hot reload Nginx to get rid of deleted open files
-echo "Hot reloading nginx..."
-docker exec nginx nginx -s reload
-
-echo "✓ All done !"

setup-scripts/blocklist-airtable.py

@@ -122,63 +122,30 @@ async def block_skylinks_from_airtable():
         )
         await send_msg(message, file=("\n".join(invalid_skylinks)))
 
-    apipassword = exec("docker exec sia cat /sia-data/apipassword")
     ipaddress = exec(
-        "docker inspect -f '{{range.NetworkSettings.Networks}}{{.IPAddress}}{{end}}' sia"
+        "docker inspect -f '{{range.NetworkSettings.Networks}}{{.IPAddress}}{{end}}' nginx"
     )
 
-    print("Sending blocklist request to siad")
+    print("Sending blocklist request to siad through nginx")
     response = requests.post(
-        "http://" + ipaddress + ":9980/skynet/blocklist",
+        "http://" + ipaddress + ":8000/skynet/blocklist",
-        auth=("", apipassword),
-        headers={"user-agent": "Sia-Agent"},
         data=json.dumps({"add": skylinks}),
     )
 
-    if response.status_code == 204:
+    print(json.dumps({"add": skylinks}))
-        print("Siad blocklist successfully updated with provided skylink")
+
-    else:
+    if response.status_code != 204:
         status_code = str(response.status_code)
         response_text = response.text or "empty response"
         message = (
-            "Siad blocklist endpoint responded with code "
+            "Airtable blocklist request responded with code "
             + status_code
             + ": "
             + response_text
         )
         return await send_msg(message, force_notify=False)
 
-    # Remove from NGINX cache
+    return await send_msg("Siad blocklist successfully updated with provided skylink")
-    # NOTE:
-    # If there are changes to how the NGINX cache is being cleared, the same
-    # changes need to be applied to the /scripts/blocklist-skylink.sh script.
-    print("Searching nginx cache for blocked files")
-    cached_files_count = 0
-    batch_size = 1000
-    for i in range(0, len(skylinks), batch_size):
-        cached_files_command = (
-            "find /data/nginx/cache/ -type f | xargs -r grep -Els '^Skynet-Skylink: ("
-            + "|".join(skylinks[i : i + batch_size])
-            + ")'"
-        )
-        cached_files_count += int(
-            exec(
-                'docker exec nginx bash -c "'
-                + cached_files_command
-                + ' | xargs -r rm -v | wc -l"'
-            )
-        )
-
-    if cached_files_count == 0:
-        return print("No nginx cached files matching blocked skylinks were found")
-    else:
-        print("Hot reloading nginx")
-        exec("docker exec nginx nginx -s reload")
-
-    message = (
-        "Purged " + str(cached_files_count) + " blocklisted files from nginx cache"
-    )
-    return await send_msg(message)
 
 
 loop = asyncio.get_event_loop()
@@ -186,6 +153,5 @@ loop.run_until_complete(run_checks())
 
 # --- BASH EQUIVALENT
 # skylinks=$(curl "https://api.airtable.com/v0/${AIRTABLE_BASE}/${AIRTABLE_TABLE}?fields%5B%5D=${AIRTABLE_FIELD}" -H "Authorization: Bearer ${AIRTABLE_KEY}" | python3 -c "import sys, json; print('[\"' + '\",\"'.join([entry['fields']['Link'] for entry in json.load(sys.stdin)['records']]) + '\"]')")
-# apipassword=$(docker exec sia cat /sia-data/apipassword)
+# ipaddress=$(docker inspect -f '{{range.NetworkSettings.Networks}}{{.IPAddress}}{{end}}' nginx)
-# ipaddress=$(docker inspect -f '{{range.NetworkSettings.Networks}}{{.IPAddress}}{{end}}' sia)
+# curl --data "{\"add\" : ${skylinks}}" "${ipaddress}:8000/skynet/blocklist"
-# curl -A "Sia-Agent" --user "":"${apipassword}" --data "{\"add\" : ${skylinks}}" "${ipaddress}:9980/skynet/blocklist"