Livecharts New - Constant Disconnects and Failed Reconnects

[dannol48]

The new service is constantly disconnecting and attempting reconnect on the main LiveCharts page. New Charts (secondary charts) are cycling too, but simply fail (no waves and dead) requiring constant toggling (change symbol as there is no refresh) to maintain connection. I'm assuming this is network time-outs or server busy problems at eSignal.

I was experiencing the problem during Beta (for months), but assumed it was due to a small service test and too many users connecting. Also, saw occassional disconnects on old LiveCharts (nothing new), but they would recover OK.

What I'm now experiencing is just plain aweful, and unusable as a service.

I'm on high-bandwidth cable internet, and have no problems with my Ameritrade Command Center or other services keeping connections including their Livecharts.

I'm also having charts flip from sub-penny to rounding and back, and dropping part day data on reconnect on day charts. I'm so busy trying to keep charts working, I'm unable to do normal tasks.

Is anyone else having the same problems?

[awuit9090]

Same things is happening here, when the new chart hangs, the new window wouldn't even close, need to shut down the main window and start all over again

[dannol48]

Hi awuit9090,

I can get the main page to reconnect using the refresh (two arrows in circle) button. However, the refresh (or reconnects) often drop the current day's data from the chart, so it only shows prior days trade range and volume on day (D) chart. There's no way to do simple refresh on that chart, other than change symbol or interval. Looks like these apps are still chocked full of problems, and NOT ready for production service.

I'm almost ready to throw in the towel. It looks like they pulled old charts now, so I'm COMPLETELY DOWN (and literally exhausted at mid-day from battling CHART-FATIGUE). I guess that is a new occupational hazard.....in same class as Carpal Tunnel. I'll be filing for dissability within a week and REFUNDS.

[dannol48]

Interesting workaround and suggestion on cause.

By using the DOW ticker $INDU for chart and tick-by-tick on the main page, it keeps service connected and a secondary chart page also remains connected and correct.

It would appear the problem may be due to a low activity timer on either the applet or the server. I note that the time clock on the applet appears to be synced with my system clock and isn't updated from the server any longer. If I'm corrent, that clock activity (one per second heartbeat) may have helped keep the old app connected.

Just a theory, but the bypass is working for the moment with the main and one chart. I'll add more back later. Oh, and high activty/trade stocks might keep it working too. Just wanted something with a steady dataflow to check first.

Someone wakeup the programmers and get them looking.

Edit: Main + Two Secondaries now working stable on bypass method. I've also switched the Main to Nasdaq $COMP, and that approx. every 15 second update also keeps connection alive.

[dannol48]

Main + 4 Charts operating and stable using the COMP$/Nasdaq on the main chart for keep-alive. I'll run this way for the day and report any problems.

Note: I'm using Windows XP with IE7 (SP2 and all current service)

Observation: Is it possible the applet now uses TCP Keep-Alive for heartbeat, but Windows OSes don't default with Keep-Alive active, i.e. you need to hack the Registry TCP attributes to turn-on Keep-Alive. I may explore that possibility for a test at a later time.

Please note, I usually get paid by large corporations for this type of work.

D. Leckington
Private Consultant
Network and Systems Integtration Engineer

[dannol48]

I'm sorry to report that all 4 secondary charts have failed. All 4 were in Day (D) mode for 4 separate Stock Symbols including DE (Deer at NYSE), so at least one had good activity prior to the failure. Day trading bar-span/activity for all 4 dropped-out and all 4 have frozen. The main chart continues to operate with COMP$. I don't know if the main chart went through a disconnect/reconnect to cause the failure.

At this point, the work-around reduces the problem, but doesn't eliminate the secondary chart failures. All other subscription services to other vendors remained 100% stable at the point of failure.

Toggling period, symbol, etc. on the secondary chart does restore it to service, but the only indication of failure being the loss of data makes it an intensitive watch process to detect the problem.

Edit1: Additional observation on restore of secondary chart after failure: The day-span range on chart does not fully restore, and appears to only reflect trades from the point of restoration.

Edit2: Running Java 6 Update 5 (build 1.6.0_05-b13) - CURRENT

Edit3: Changing Stock symbols on a failed secondary chart and back to prior symbol does do a full restore of data and the chart then accurately reflects the full day-span. My prior restore test in Edit1 was simply using a change of period on same symbol. I'm guessing that changing symbol forces a full data refresh from the server. So, this method prevents having to drop and restart the failed secondary chart (unless I'm missing something).

[dannol48]

Java Console shows following on session failure, Main reconnect and all 4 secondaries fail with no recovery. CSCO used for chart heartbeat on Main.

New session state: WaitingBetweenConnections
10:25:20 Communication error: java.net.SocketException: Resource temporarily unavailable: recv failed
Sender thread (S) #1 exiting.
Socket (S) #1 closed.
10:25:31 Attempting to connect.
Server connection (S) #2 created.
New session state: Communication, connected = false
10:25:31 Socket connected
Client Protocol # = 232
IMAF is disabled (Invisible Off).
Server Name = LC-H-03
Server URL = lc-h-03.esignal.com
User authorized.

[dannol48]

Speed Test Report from Speakeasy (for Engineers @ eSignal)

Using Roadrunner Cable Internet from Columbus, OH

Location Download* Upload*
Chicago 6575 472
Atlanta 4636 477
NY 5913 478
Dallas 3649 470
DC 5825 470
Seattle 3204 472
LA 2906 473
SF 3169 470

* Kbps

[dannol48]

New problem. App is now spiking 100% system use.

Only difference this morning, I'm running Java Console (changed from Hide to Show Console in Adanced settings) to see if there was any beneficial information at app launch), which I usually don't do. Unknown, if that change is cause or coincidental.

Oh...and I'm using a high activity stock CSCO in the Main. That's a switch from previously using DOW or Nasdaq for keep-alive heartbeat.

Edit1: Trading System is Athlon 64 2800+ w/1GB Dual-Channel DDR400 in operation about 3+ years.

Edit2: I'm assuming today's 100% processor sustained spikes are the problem previously reported by other Forum members. I also had to cancel the LC Task to end the session. I'm also assuming the use of high-data rate symbols, like Cisco (CSCO), is a possilbe cause. This leaves me with no apparent method to both stabilize chart sessions and stabilize the system, as low data rates drop chart sessions and high data rates are contributing to system failures. At best, I'm getting about 1-2 hours of use between failures.

[eSignal_ToddI]

Thanks dannol48 for your contributions on this issue.

I have forwarded this information as well to engineering as they continue their investigation on this issue.