Re: [Quickfixj-users] Production issue - Loss of messages + inaccurate time

[Christoph J. <chr...@ma...>]

What happens after you re-logon? Doesn't the counterparty send a resend request and the "lost" order is resent?
To me this sounds like a network problem of some sort. Your side thinks the connecting is still up while the other side probably waits for your heartbeats, not realising that the connecting is broken.
If this happens in times where no messages are sent then maybe there is an idle timeout somewhere that kills the connection? I think your network guys should monitor the connection and be able to find out what is going on by taking tcpdumps or similar.
Cheers
Chris

________________________________
Von: Elyse Badr <ely...@ca...>
Gesendet: Donnerstag, 18. Mai 2023 08:23
An: qui...@li... <qui...@li...>
Betreff: Re: [Quickfixj-users] Production issue - Loss of messages + inaccurate time

QFJ Documentation: https://che01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.quickfixj.org%2Fdocumentation%2F&data=05%7C01%7Cchristoph.john%40macd.com%7C37ca9700b2124e0c20c508db5768980f%7Cfb673ffa32b24d71b08c8c56f99c56ca%7C0%7C0%7C638199878965850633%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=Dze1ZgeB6E95FWa%2BXXPv3HNnSGwipXjqom4s6OC3kS8%3D&reserved=0<http://www.quickfixj.org/documentation/>
QFJ Support: https://che01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.quickfixj.org%2Fsupport%2F&data=05%7C01%7Cchristoph.john%40macd.com%7C37ca9700b2124e0c20c508db5768980f%7Cfb673ffa32b24d71b08c8c56f99c56ca%7C0%7C0%7C638199878965850633%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=gMcEHwTfhsKYWQrvti07VHSUdPJxaFg5ZU80N3jXhNw%3D&reserved=0<http://www.quickfixj.org/support/>


Hi all,

The delay in messages issue was solved (load on server). But the time problem is not, it is occurring on 2 different servers even if the servers time is in sync with the internet. The weird behavior, is that when no market data / info securities is received from an amount of time:
-  the heartbeat messages are not received - (note that the last heart beat message received from market is 2 seconds lower than our last heartbeat)
- a test request is sent, no reply from market
- if an order is sent, it is lost and does not reach the market
- a disconnection occurs
- a logon is sent again
- The market replies - so the market only replies after a logon is initiated

Do you think that the 2 seconds differences cause the market to disconnect from our session until log on again? If yes why don't we receive the disconnection message from the market side?
And can we solve similar issues by changing in the session config instead of changing the server time ? (to prevent this behavior on prod for example)




_______________________________________________
Quickfixj-users mailing list
Qui...@li...
https://che01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.sourceforge.net%2Flists%2Flistinfo%2Fquickfixj-users&data=05%7C01%7Cchristoph.john%40macd.com%7C37ca9700b2124e0c20c508db5768980f%7Cfb673ffa32b24d71b08c8c56f99c56ca%7C0%7C0%7C638199878965850633%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=nxFF4oYJuZebwlewYKsZH%2Bddf7PotddVdnU%2FLJfmdxs%3D&reserved=0<https://lists.sourceforge.net/lists/listinfo/quickfixj-users>