Re: [Quickfix-developers] concurrent threads -- "External component hasthrown an exception"
Brought to you by:
orenmnero
Menu
▾
▴
Re: [Quickfix-developers] concurrent threads -- "External component hasthrown an exception"
|
From: Rick L. <ric...@gm...> - 2008-06-25 16:14:09
|
It is very /not /OK to have nested for-loops w/ the same counter
variable -- just another side effect of my cut-paste job. Sorry for the
confusion....
Rick
John Haldi wrote:
> I'm not a C# guy, but is it ok to have nested For loops with the same
> counter variable? In VB this would be dangerous:
>
> For i = 0 to .....
> For i = 0 to ......
>
> Next
> Next
>
> ------------------------------------------------------------------------
> *From:* qui...@li...
> [mailto:qui...@li...] *On Behalf
> Of *Rick Lane
> *Sent:* Wednesday, June 25, 2008 11:34 AM
> *To:* qui...@li...
> *Subject:* [Quickfix-developers] concurrent threads -- "External
> component hasthrown an exception"
>
> Greetings,
>
> I have provided a slimmed-down version of the two threads that are
> running concurrently in my application in an attempt to see if anyone
> can pinpoint why I might be getting this ugly exception when grabbing
> integer values from QuickFix.Message objects.
>
> The first thread is my "producer" thread, and it basically converts
> FAST data into a valid FIX message, constructs a QuickFix.Message from
> this FIX string and then adds this Message object onto the "consumers"
> queue.
>
> The second (consumer) thread waits for new QuickFix.Message objects to
> be added to the queue, and then does the necessary processing.
>
> When the exception occurs, I print out the FIX message text in
> question to see if there is something errant about it -- I then, in a
> separate application, construct a QuickFix.Message object from the
> /same /FIX message string (that caused the original exception) and it
> works fine -- so it's definitely not a FIX-formatting issue. It must
> be a thread issue because the bug is very intermittent, some machines
> it happens every few minutes, while some machines it happens only once
> or twice a week.
>
> Please let me know if you see any glaring issues -- the same
> QuickFix.Message object is never accessed on more than one thread at a
> time, so it seems that there must be some static or shared component
> of QuickFix that is having concurrency issues when 2 threads are
> accessing it at the same time (meaning, I might have 2
> QuickFix.Message objects, one on each thread, and each thread is
> performing actions on these objects individually -- which should be
> allowed, but if each of these distinct objects has some shared
> component, then I can see a problem arising).
>
> Thanks in advance!
>
> Rick
>
> private void runProducerThread()
> {
> while (WaitHandle.WaitAny(m_SyncEventsMarketData.EventArray) != 1) {
> // raw FAST-decoded data will come in through this Queue
> ArrayList newMessages = new ArrayList();
> lock (m_QueueMarketData.SyncRoot) {
> newMessages.AddRange(m_QueueMarketData);
> m_QueueMarketData.Clear();
> }
>
> for (int msgNum = 0; msgNum < newMessages.Count; msgNum++) {
> FASTDataPacket packet = (FASTDataPacket)newMessages[msgNum];
> // my MarketDataDecoder will convert the FAST-encoded data
> into a valid FIX message string
> string msgText = m_MarketDataDecoder.ProcessPacket(packet);
> // I create a QuickFix.Message object with this string and
> my DataDictionary
> QuickFix.Message message = new QuickFix.Message(msgText,
> m_DataDictionary);
> int msgSeqNum = message.getHeader().getInt(34);* *
>
> if (msgSeqNum = m_NextMarketDataSeqNum) {
> // The other (consumer) thread, which I'm adding this
> QuickFix.Message to will do the actual
> // processing of the *contents* of the message -- this
> thread simply ensures that we don't
> // miss any UDP packets by looking at the MsgSeqNum of
> each incoming packet. The other
> // thread, then, doesn't have to worry about this.
> lock (m_ConsumerQueue.SyncRoot) {
> m_ConsumerQueue.Enqueue(message);
> m_ConsumerSyncEvents.NewItemEvent.Set();
> }
> m_NextMarketDataSeqNum = msgSeqNum + 1;
> } else {
> // reconcile missed messages...
> }
> }
> }
> }
> private void runConsumerThread()
> {
> // Wait for producer thread to add another QuickFix.Message to my Queue
> while (!m_ConsumerSyncEvents.ExitThreadEvent.WaitOne(0, false) &&
> WaitHandle.WaitAny(m_ConsumerSyncEvents.EventArray) != 1) {
> ArrayList newMessages = new ArrayList();
>
> lock (m_ConsumerQueue.SyncRoot) {
> newMessages.AddRange(m_ConsumerQueue);
> m_ConsumerQueue.Clear();
> }
> for (int i = 0; i < newMessages.Count; i++) {
> // Perform necessary processing on this Message object
> QuickFix.Message message = (QuickFix.Message)newMessages[i];
> int msgSeqNum = message.getHeader().getInt(34);
> int numEntries = message.getInt(268);
> // Each update will have multiple MDEntries
> for (uint i = 0; i < numEntries; i++) {
> QuickFix44.MarketDataIncrementalRefresh.NoMDEntries
> group = new QuickFix44.MarketDataIncrementalRefresh.NoMDEntries();
> message.getGroup(i + 1, group);
> string securityID = group.getSecurityID().getValue();
> MDPInstrument inst =
> (MDPInstrument)m_ISINToInstrumentMap[securityID];
>
> if (inst != null) {
> if (group.isSetField(83)) {
> int rptSeqNum = group.getField(new
> IntField(83)).getValue();
> if (inst.lastMarketDataSeqNum != -1 &&
> msgSeqNum > inst.lastMarketDataSeqNum) {
> // process inc refresh
> MDUpdateAction updateAction =
> group.getMDUpdateAction();
> int entrySize = 0;
> if (group.isSetMDEntrySize()) {
> MDEntrySize mdEntrySize =
> group.getMDEntrySize();
> entrySize = (int)mdEntrySize.getValue();
> }
> double entryPrice = 0;
>
> MDEntryPx entryPx = group.getMDEntryPx();
> entryPrice = entryPx.getValue();
>
> MDEntryType entryType =
> group.getMDEntryType();
> bool implied = false;
>
> QuoteCondition quoteCondition = null;
> if (group.isSetQuoteCondition()) {
> quoteCondition =
> group.getQuoteCondition();
>
> if (group.isSetQuoteCondition()) {
> if
> (group.getQuoteCondition().getValue()[0] == 'K') {
> implied = true;
> }
> }
> }
> if (group.isSetField(1020)) {
> int tradeVolume = group.getField(new
> IntField(1020)).getValue();
> inst.SetTradeVolume(tradeVolume);
> }
>
> if ((quoteCondition == null ||
> quoteCondition.getValue()[0] != QuoteCondition.EXCHANGE_BEST)) {
> int priceLevel = -1;
> switch (entryType.getValue()) {
> case MDEntryType.BID: {
> // This is where the exception is occurring
> *priceLevel =
> group.getField(new IntField(1023)).getValue();** *
> switch
> (updateAction.getValue()) {
> case MDUpdateAction.NEW:
> // do necessary
> processing...
> break;
> case
> MDUpdateAction.DELETE:
> // do necessary
> processing...
> break;
> case
> MDUpdateAction.CHANGE:
> // do necessary
> processing...
> break;
> }
> }
> break;
> case MDEntryType.OFFER: {
> // This is also where the exception is
> occurring
> *priceLevel =
> group.getField(new IntField(1023)).getValue();** *
> switch
> (updateAction.getValue()) {
> case MDUpdateAction.NEW:
> // do necessary
> processing...
> break;
> case
> MDUpdateAction.DELETE:
> // do necessary
> processing...
> break;
> case
> MDUpdateAction.CHANGE:
> // do necessary
> processing...
> break;
> }
> }
> break;
> case MDEntryType.TRADE_VOLUME:
> // do necessary processing...
> break;
> case
> MDEntryType.TRADING_SESSION_LOW_PRICE:
> // do necessary processing...
> break;
> case
> MDEntryType.TRADING_SESSION_HIGH_PRICE:
> // do necessary processing...
> break;
> case MDEntryType.SETTLEMENT_PRICE:
> // do necessary processing...
> break;
> }
> }
> }
> }
> }
> }
> }
> }
> }
>
> No virus found in this incoming message.
> Checked by AVG.
> Version: 8.0.101 / Virus Database: 270.4.1/1518 - Release Date:
> 6/25/2008 9:46 AM
>
|
