I did see in the forum that there is another question about the message Write failed: EDC5140I Broken pipe. (errno2=0x76697242). We get the same error on an intermittent bases. When we get this error, zos is the client doing multiple gets on multiple files.
I’m wondering if key word ServerAliveInterval would help by adding it to /etc/ssh/ssh_config.
Thanks Matt
From the z/OS OpenSSH User's Guide zos 2.2
ServerAliveInterval - Sets a timeout interval in seconds after which if no data has been received from the server, ssh sends a message through the encrypted channel to request a response from the server. The default is 0, indicating that these messages are not sent to the server. Restriction: This option applies to protocol version 2 only.
tÝI¨: Opening dataset DD:DD01 for write
tÝI¨: Closing dataset //P.PWHSTD55.WHST52.FEDEXTRK - 139380 bytes received, 345
gsÝI¨: Transfer options: clientcp=IBM-1047,lrecl=404,mode=text,mount,recfm=fb,se
tÝI¨: Opening dataset DD:DD01 for write
Write failed: EDC5140I Broken pipe. (errno2=0x76697242)
Broken Pipe Error
Re: Broken Pipe Error
This error ("Write failed: EDC5140I Broken pipe. (errno2=0x76697242)"), refers to the pipe connection between Co:Z SFTP and the IBM ssh client process being lost. This would occur if the ssh child process terminated for some reason.
It seems unlikely that ServerAliveInterval would help, since this has to do with retries when nothing is happening on the connection, which is not the case in a Co:Z SFTP batch job.
If this is reproducible, then turn on "-vvv" debug tracing so that ssh prints a trace of what is happening. This might point to the reason why it is failing/terminating.
It seems unlikely that ServerAliveInterval would help, since this has to do with retries when nothing is happening on the connection, which is not the case in a Co:Z SFTP batch job.
If this is reproducible, then turn on "-vvv" debug tracing so that ssh prints a trace of what is happening. This might point to the reason why it is failing/terminating.
Re: Broken Pipe Error
So I finally got a fix in place for this issue, I had to open a pmr with IBM and I provided many traces to them. I think the right IBM'r got involved and finally made a suggestion which has corrected this problem. I needed to apply PTF UI62081 for zos 2.2 and add SELECTIVEACK TCPTIME to the TCPCONFIG statement. We are now good to go with connecting this particular server.
Wanted to add this to the post in hopes of helping others. Matt
Wanted to add this to the post in hopes of helping others. Matt
Re: Broken Pipe Error
Interesting, so the server you were communicating with supported SACK and was experiencing significant packet loss (some times), which uncovered a bug in the Comm Server TCP/IP stack?