Implement retry on failed sends

chaind_eth/data/config/config.ini

@@ -3,6 +3,7 @@ socket_path =
 runtime_dir =
 id =
 data_dir =
+dispatch_delay = 4.0
 
 [database]
 engine =

chaind_eth/dispatch.py

@@ -56,11 +56,13 @@ class Dispatcher:
                 logg.debug('too many inflight txs for {}, skipping {}'.format(sender, k))
                 continue
             logg.debug('processing tx {} {}'.format(k, txs[k]))
+            r = 0
             try:
-                self.adapter.dispatch(self.chain_spec, rpc, k, txs[k], session)
+                r = self.adapter.dispatch(self.chain_spec, rpc, k, txs[k], session)
             except JSONRPCException as e:
                 logg.error('dispatch failed for {}: {}'.format(k, e))
                 continue
+            if r == 0:
                 self.inc_count(sender, session)
                 c += 1
         return c

chaind_eth/runnable/server.py

@@ -35,11 +35,13 @@ argparser = chainlib.eth.cli.ArgumentParser(arg_flags)
 argparser.add_argument('--data-dir', type=str, help='data directory')
 argparser.add_argument('--runtime-dir', type=str, help='runtime directory')
 argparser.add_argument('--session-id', dest='session_id', type=str, help='session identifier')
+argparser.add_argument('--dispatch-delay', dest='dispatch_delay', type=float, help='socket timeout before processing queue')
 args = argparser.parse_args()
 extra_args = {
     'runtime_dir': 'SESSION_RUNTIME_DIR',
     'data_dir': 'SESSION_DATA_DIR',
     'session_id': 'SESSION_ID', 
+    'dispatch_delay': 'SESSION_DISPATCH_DELAY',
         }
 #config = chainlib.eth.cli.Config.from_args(args, arg_flags, default_config_dir=config_dir, extend_base_config_dir=config_dir)
 config = chainlib.eth.cli.Config.from_args(args, arg_flags, extra_args=extra_args, base_config_dir=config_dir)
@@ -56,7 +58,8 @@ if not config.get('SESSION_SOCKET_PATH'):
     config.add(socket_path, 'SESSION_SOCKET_PATH', True)
 
 if config.get('DATABASE_ENGINE') == 'sqlite':
-    config.add(os.path.join(config.get('SESSION_DATA_DIR'), config.get('DATABASE_NAME') + '.sqlite'), 'DATABASE_NAME', exists_ok=True)
+    #config.add(os.path.join(config.get('SESSION_DATA_DIR'), config.get('DATABASE_NAME') + '.sqlite'), 'DATABASE_NAME', exists_ok=True)
+    config.add(os.path.join(config.get('SESSION_DATA_DIR'), config.get('DATABASE_NAME')), 'DATABASE_NAME', exists_ok=True)
     
 config.censor('PASSWORD', 'DATABASE')
 logg.debug('config loaded:\n{}'.format(config))
@@ -83,7 +86,8 @@ class SessionController:
         self.srv = socket.socket(family=socket.AF_UNIX, type=socket.SOCK_STREAM)
         self.srv.bind(config.get('SESSION_SOCKET_PATH'))
         self.srv.listen(2)
-        self.srv.settimeout(4.0)
+        self.srv.settimeout(float(config.get('SESSION_DISPATCH_DELAY')))
+
 
     def shutdown(self, signo, frame):
         if self.dead:

chaind_eth/runnable/syncer.py

@@ -64,7 +64,7 @@ if not config.get('SESSION_SOCKET_PATH'):
     config.add(socket_path, 'SESSION_SOCKET_PATH', True)
 
 if config.get('DATABASE_ENGINE') == 'sqlite':
-    config.add(os.path.join(config.get('SESSION_DATA_DIR'), config.get('DATABASE_NAME') + '.sqlite'), 'DATABASE_NAME', exists_ok=True)
+    config.add(os.path.join(config.get('SESSION_DATA_DIR'), config.get('DATABASE_NAME')), 'DATABASE_NAME', exists_ok=True)
     
 config.censor('PASSWORD', 'DATABASE')
 logg.debug('config loaded:\n{}'.format(config))

chainqueue/adapters/eth.py

@@ -1,5 +1,6 @@
 # standard imports
 import logging
+import datetime
 
 # external imports
 from chainlib.eth.constant import ZERO_ADDRESS
@@ -11,13 +12,15 @@ from hexathon import (
         add_0x,
         strip_0x,
         )
-from chainqueue.enum import StatusBits
+from chainqueue.enum import (
+        StatusBits,
+        errors as queue_errors,
+        )
 
 # local imports
 from chainqueue.adapters.base import Adapter
 
-#logg = logging.getLogger(__name__)
+logg = logging.getLogger(__name__)
-logg = logging.getLogger()
 
 
 class EthAdapter(Adapter):
@@ -39,7 +42,12 @@ class EthAdapter(Adapter):
 
 
     def upcoming(self, chain_spec, session=None):
-        return self.backend.get(chain_spec, StatusBits.QUEUED, self.translate) # possible maldesign, up-stack should use our session?
+        txs = self.backend.get(chain_spec, self.translate, session=session, status=StatusBits.QUEUED, not_status=StatusBits.IN_NETWORK)
+        before = datetime.datetime.utcnow() - self.error_retry_threshold
+        errored_txs = self.backend.get(chain_spec, self.translate, session=session, status=StatusBits.LOCAL_ERROR, not_status=StatusBits.FINAL, before=before, requeue=True)
+        for tx_hash in errored_txs.keys():
+            txs[tx_hash] = errored_txs[tx_hash]
+        return txs
 
 
     def add(self, bytecode, chain_spec, session=None):

requirements.txt

@@ -1,3 +1,3 @@
-chaind<=0.0.1,>=0.0.1a8
+chaind<=0.0.2,>=0.0.2a3
 hexathon~=0.0.1a8
 chainlib-eth<=0.1.0,>=0.0.9a4

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = chaind-eth
-version = 0.0.1a6
+version = 0.0.2a1
 description = Queue server for ethereum
 author = Louis Holbrook
 author_email = dev@holbrook.no