Current location | Thread information | |
![]() ![]() ![]() ![]() ![]() ![]() |
Last Activity 2/20/2018 9:38 AM 52 replies, 4769 viewings |
|
Printer friendly version |
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Hi, Brian. I see each TP's log files. From each TP's View->LogFiles drop down, I see: Local Disk(C:)>ProgramData>NirvanaSystems>TradeProcessor>Logs>MyOVID>MyOVAcctName>2016>Jul. In this folder, I see the following folders: ErrorLogs Reports and Alerts Strategies Symbols TP NOTE: ProgramData is (normally) a hidden folder. Is that why you can't find Logs through FileExplorer? | ||
^ Top | |||
Barry Cohen![]() Icon ![]() ![]() ![]() ![]() ![]() ![]() Posts: 1844 Joined: 10/11/2012 ![]() |
In Windows Server 2008, those files are located at: C:\Users\Public\Nirvana Systems\Trade Processor\Logs. The View Logs function in TP currently isn't able to get there. | ||
^ Top | |||
Mark S![]() Regular ![]() ![]() Posts: 56 Joined: 10/11/2012 Location: Barrington, IL ![]() |
I spoke to soon saying that my 3 TP/IBG instances were running flawlessly. On Monday of this week the closing orders were not executed properly on several securities. Closing them out today cost me several thousand dollars compared to what it would have been on Monday. I am not very technical, but am wondering what I should be looking to correct so this doesn't happen again. Any ideas? Did anyone else have problems on Monday 7/25? | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
This past Monday, I started a 3rd TP/IBG pair. Since then, I did see one disconnect occur (I think it was an IBG connection that was lost). This morning, I started a 4th TP/IBG pair. And over the next hour, I had a couple IBG connections drop during the critical minutes before and after the market opening. I'm running this on a "t2.micro" instance. Over the weekend, I will transition to a larger instance. This t2.micro AWS server appears to be running too close to the edge for reliable service with 4 TP/IBG pairs, and perhaps 3 pairs is also too close to the edge. [Edited by SteveL on 7/30/2016 9:02 AM] | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
Mark: I did have problems Monday but not TP/IB gateway/execution issues. Problems were because my server rolled over and died due to lack of storage (fortunately Monday AM trades were submitted before server blew up). I think logging of Monday's trades dealt my server's last hand. However, I had no execution problems. After I created a new AWS instance and re-loading TP/IBG, all OV positions were reconnected without problem too. BTW, building a new AWS instance is not the only way to address lack of disk capacity (e.g. delete files, etc.) but that's the avenue I chose for a clean platform because I'm surprised the capacity was used, so I can now watch to assure the log file capacity is released. I case you don't know, to check disk capacity, open server's Windows Explorer, right click on your "Local disk (C:)" and click on Properties. If you are out of "Free Space", that's what has caused my server failure. Otherwise, Steve's point of RAM may be something to look at too. Also, I'm running 3 TP/IBG. Have run as many as 4 but not recently. And I am keeping TP Logging to 1 week in each TP. [Edited by BrianD on 7/29/2016 5:09 PM] | ||
^ Top | |||
Mark S![]() Regular ![]() ![]() Posts: 56 Joined: 10/11/2012 Location: Barrington, IL ![]() |
Steve, where do you look to determine if a connection was lost? In looking at the AWS monitoring capabilities, it looks like the CPU is operating at 3-5% normally, with spikes to 10-15% capacity, but I don't see anything that looks like a connections monitor. On 7/25, when I missed several closing orders, everything looked like it was operating normally from AWS. I'm running 3 instances on a micro, and trying to determine if I need to upgrade to a small. I do see that I have less than 1GB of storage left. [Edited by Mark S on 7/30/2016 8:18 AM] | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Hi MarkS. I got a popup telling me the connection was lost. I think it was a TP popup. I'll try to remember to snapshot future popups and post them. I've also gotten IBG popups when a connection is lost. I had a "connection failed" popup from IBG this morning because the newest accounts were connected to the Master/Advisor account and trading was disabled by IB on those client logins (which I've now re-enabled from the Master account). NOTE: IB makes those changes on Friday night after market close. Windows Task Manager is where you see CPU and memory status. Here is a snapshot showing where you open the Task Manager: After Task Manager is opened, select the Performance tab, and select the radio button for the resource you want to see. I've selected Memory: The pagefile is used by the OS to handle over committed memory needs. So, if the pagefile is big enough, then I think, theoretically, things should continue to work. As memory requirements exceed the 1GB available on a micro server, then paging starts happening (things get swapped in and out of the 1GB RAM memory to the pagefile on some storage device which is slower than RAM). You'll see in this snapshot that the 1GB is overallocated. The pagefile has 1.9GB of committed memory. But, there is still plenty of pagefile space left. I suspect that TP and IBG are designed with timeouts and expectations about how fast things like connections should be established. So, when you have too many things running, then the CPU work queue starts getting longer, and things start timing out. That is my interpretation of what happened on Friday morning. So, the problems I saw could have been due to 1 CPU not keeping up, or the pagefile access time is not fast enough. If my analysis is correct, then either a 2nd CPU or more memory will resolve the connection reliability problem. Unfortunately, to get a 2CPU AWS server, you have to move up 2 levels. t2.Micro has 1 CPU and 1GB. t2.Small has 1CPU and 2GB. t2.Medium has 2 CPUs and 4GB. Each step doubles the cost. So, I'll try the first step and see how it goes. In the grand scheme of things, these costs are minimal, and worth avoiding a missed trade entry/exit. [Edited by SteveL on 7/30/2016 9:00 AM] ![]() ![]() | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
MarkS. I see you've edited your post since I started working on my response. So, obviously you already knew some of what I posted. I believe the CPU usage is MUCH higher (close to 100%) around the times that are configured for the TPs to start doing things (like placing trades). Which makes me think that one other solution might be to set different times for each of the TPs - stagger them 1 minute apart for placing trades. I think I'll try that first, and see how things go. | ||
^ Top | |||
Mark S![]() Regular ![]() ![]() Posts: 56 Joined: 10/11/2012 Location: Barrington, IL ![]() |
Steve, thank you! I ended up launching a new small instance of AWS and installing the new TP4 and associated IBG with it. I have 4GB of the 30GB left. For fun and backup I just stopped the micro instance, and compressed the hard drive, which made it go from 11MB to 1GB of available space. I had no idea about the Task Manager capabilities - thank you! I'll try to watch that at 8:30 Monday morning. I thought the idea of staggering trade times made a lot of sense, so I've done that, too. | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
I do not think CPU is the biggest factor and staggering should help. But, RAM certainly can be. Developers get used to >4GB RAM for appl. development, and so little optimization. Also, I doubt there is a lot of computation going on, mostly network and graphics? Would not count on paging to address lack of RAM in windows. I suspect physical RAM and disk space are the things to watch most. Also, monitors eat up capacity. | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
This morning, I have the 4 TP/IBG pairs staggered in their order submission times. The 3rd pair failed with this error (see below), although the orders were placed. I restarted TP, and all are currently connected. The one unique thing I can think of about the failing TP is that it placed both an opening order as well as closing orders. The others only placed closing orders. NOTE: I have the submission times staggered by 5 minutes. [Edited by SteveL on 8/1/2016 8:22 AM] ![]() | ||
^ Top | |||
Mark G![]() Member ![]() Posts: 41 Joined: 10/11/2012 ![]() |
Yup, that is the same exact error message I have been getting . . . I've just given up with this whole mess. --Mark G. | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
I have 3 running this AM. All opens and closes. Ran fine. Sounds like a bug though. | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Again this morning, the SAME ACCOUNT TradeProcessor failed. That made me focus on what might be unique about this account. It has been recently moved to IB. It contains 1 share of an old worthless symbol, which TP (I assume because I've configured that way) tried to close. I'll have to work with IB to get that worthless item out of the account. So otherwise, all the other accounts are running fine on a t2.micro. Here were the two errors produced by this failing TP: Problem signature: Problem Event Name: APPCRASH Application Name: TradeProcessor.exe Application Version: 1.4.20.0 Application Timestamp: 573a408c Fault Module Name: KERNELBASE.dll Fault Module Version: 6.3.9600.18340 Fault Module Timestamp: 5736541b Exception Code: e0434352 Exception Offset: 00014878 OS Version: 6.3.9600.2.0.0.272.7 Locale ID: 1033 Additional Information 1: 0ea4 Additional Information 2: 0ea4d10b75a2cc218ace190122c26039 Additional Information 3: b213 Additional Information 4: b213103fad9442fa10ddc5231e319139 Read our privacy statement online: http://go.microsoft.com/fwlink/?linkid=280262 If the online privacy statement is not available, please read our privacy statement offline: C:\Windows\system32\en-US\erofflps.txt Problem signature: Problem Event Name: APPCRASH Application Name: TradeProcessor.exe Application Version: 1.4.20.0 Application Timestamp: 573a408c Fault Module Name: KERNELBASE.dll Fault Module Version: 6.3.9600.18340 Fault Module Timestamp: 5736541b Exception Code: c000041d Exception Offset: 00014878 OS Version: 6.3.9600.2.0.0.272.7 Locale ID: 1033 Additional Information 1: 5faa Additional Information 2: 5faa46876913924e5bf842e8d4040f41 Additional Information 3: 7451 Additional Information 4: 7451ecef86d567f04c612b2fccba3fa8 Read our privacy statement online: http://go.microsoft.com/fwlink/?linkid=280262 If the online privacy statement is not available, please read our privacy statement offline: C:\Windows\system32\en-US\erofflps.txt | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
Steve: Did you send this to Support? | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Brian: No, I have not sent anything to support. I assume (and pretty sure) that Barry monitors this forum, and will ask for more info if needed. I think the cause of all the failures this past week have been due to that untradeable stock that was a leftover from long ago. I agree with your implication that TP should handle this without crashing. The necessary form has been submitted to IB to remove the symbol (that is no longer tradable) from the account. So, I expect that either tonight or tomorrow it will be removed, and I'll see if the failures stop. NOTE: As of this morning, I'm now running 5 TP/IBG pairs on this t2.micro. | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Sometime during the night an IBG connection dropped on one account. Had to stop/restart IBG to get it reconnected. NOTE: I can see that I need to specify separate directories for each IBG instance to keep their logs, etc. separate. Possibly/probably an indication that I need a bigger AWS server. NOTE: "Committed Memory" is 2.7GB (on a t2.micro with 1GB of RAM). Also, the previously failing TP again failed when submitting orders because (I assume) the no-longer-tradeable symbol is still in the account. Problem signature: Problem Event Name: CLR20r3 Problem Signature 01: TradeProcessor.exe Problem Signature 02: 1.4.20.0 Problem Signature 03: 573a408c Problem Signature 04: mscorlib Problem Signature 05: 4.6.1055.0 Problem Signature 06: 563c0eac Problem Signature 07: 3fd2 Problem Signature 08: d Problem Signature 09: N3CTRYE2KN3C34SGL4ZQYRBFTE4M13NB OS Version: 6.3.9600.2.0.0.272.7 Locale ID: 1033 Additional Information 1: 5861 Additional Information 2: 5861822e1919d7c014bbb064c64908b2 Additional Information 3: 6ee2 Additional Information 4: 6ee2432e64e99bdcea5e4be767a3b1a2 Read our privacy statement online: http://go.microsoft.com/fwlink/?linkid=280262 If the online privacy statement is not available, please read our privacy statement offline: C:\Windows\system32\en-US\erofflps.txt | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Last Wednesday after the prior connection failure, I moved from the AWS t2.micro to a t2.small server. And over the weekend, IB removed the 1 sh of an untradeable asset from one of the accounts. Since moving to the t2.small (3 trading days now), 5 TP/IBG pairs have not had any IB connection errors (which had occurred several times on the t2.micro with 4 pairs). And, this morning, with the untradeable asset removed from the one repeatedly failing account, the TP error did not recur. So, I'm expecting smooth sailing going forward on the t2.small. The one remaining problem is TP (pre-release 4D) does not handle delayed opening orders. It submits the order, gets the response from IB that cash is not available (because the close orders have not yet executed), but apparently never reissues the failed opening orders. Manual intervention is required to place the failed orders after cash has been freed by the close orders. | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
I have had same problem with delayed orders. | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
I spoke with Barry about it at the BASH. I got the impression it would not be a priority item until there were a more people experiencing the problem. | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
It just does not delay orders. I've been running at 9:30:45 assuming IB would be run from web page shortly, but... 9:30:45 has worked fine as a circumvention. | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Brian, There are 4 times: Between xx:xx:xx and xx:xx:xx Delay opening orders xx:xx:xx Close Orphaned Trades at xx:xx:xx How do you have these configured? Is it: Between 9:30:45 AM and 4:00:00 PM Delay opening orders NOT SELECTED? Thanks. | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
I also meant to ask: How does that work with submitting both close and opening orders together when the closing orders need to happen first? | ||
^ Top | |||
BrianD![]() Legend ![]() ![]() ![]() Posts: 302 Joined: 2/23/2013 Location: Grand Rapids, MI ![]() |
Yes to first post. To second post, Been working fine Opening and closing positions at same time, so far 😊 | ||
^ Top | |||
SteveL![]() Veteran ![]() ![]() ![]() ![]() Posts: 189 Joined: 10/11/2012 Location: Boulder, CO ![]() |
Thanks. One more question: Is this an IRA account where the close and open are being submitted at the same time? I'm surprised it works. Maybe (?) the close orders are submitted first by TP, and those close orders have actually executed by the time the opening orders are submitted (perhaps a few milliseconds later?). |
|
Legend | Action | Notification | |||
Administrator
Forum Moderator |
Registered User
Unregistered User |
![]() |
Toggle e-mail notification |