When users client sessions fail and as a result are disconnected from the database how would you proceed? Following are some example errors. Others report being disconnected without specifying the error. OpenEdge 11.5.1 64-bit DB, 32-bit clients. Client networking. Windows PC’s and database server.
From prior experience I suspect the OS and network are not responding to a required resource. At this point I am looking for a tool or technique that will expose the issue. From the OS and Network team they see no issues with other applications. They see no evidence of any network or hardware issues not related to this application. The conclusion is that the issue is with the application and OpenEdge.
Incomplete write when writing to the server (735)
Error reading socket, ret = 10054, errno=2. (778)
SYSTEM ERROR: I/O error 22 in readit, ret -1, file 8(z:\DLC\gui\adecomm.pl), addr 40512399278604288. (290)
Error reading socket=1484 ret=-1 errno=10053. (795)
errno=22 reading promsgs file, it may have been deleted.
Given both the (735) and the (290), my first guess would be a network issue of some kind between client and server. I'd run Wireshark or tshark on both the client machine and the database server, each filtering on traffic to/from the other, and wait for the errors to happen again. You can capture to a ring buffer so you don't have to worry about continually-growing capture files if it takes a while.
Then stop the captures once the error happens and compare them. You'd be looking for evidence of packet loss, e.g. many duplicate ACKs or fast TCP retransmissions. Also, a packet that appears in one trace but not the other clearly indicates packet loss. It's a bit tedious and time-consuming, and may require someone with networking expertise, but that's the approach I'd take.
Another option is to run a client server-side. It isn't an option to resolve the problem of course, but it could help to demonstrate to the non-believers that the issue isn't OpenEdge or the application. If a server-side TCP client runs fine but a remote TCP client fails with errors, that fairly clearly points to either a networking problem on the client or between the client and the server.
Also, it looks like you're reading a .pl over the network. Are you running the client with -pls?
Thank you for the input. I will need to follow up on the start up values. My best guess is that -pls is not set.
Are you sure that there are no virus scanners running on the db server?
I do not know the virus scanner configuration. Are you asking if the DB files are omitted or if there is any scanning at all?