We noticed these entires in our errlog for the first time today - where could we find some more info on what's happened here?
Is it just indicating that the unirpc service crashed? We're running UV 12/RHEL 7.9
Mon Jul 25 07:17:09 -17 root /usr/uv/bin unirpc: No Connection
Mon Jul 25 07:17:09 -17 root /usr/uv/bin uvnetd: problem send first packet.
------------------------------
Matthew Wright
Analyst
Tomago Aluminium Co PTY Ltd.
Tomago NSW AU
------------------------------
Matthew,
Difficult to tell in isolation and unless easily repeatable so it can be monitored, but the most likely candidates are all centred around a breakdown in the data exchange between client and server processes, including:
- The TCP connection between the client and server processes was abruptly terminated at the network level
- The client process unexpectedly terminated - if this is the case then there may be entries n the client system error log files.
- JAVA resource tuning or web server configuration can be factors here.
- The server process unexpectedly terminated - if this is the case then look at the server log files and for possible core files being generated at the time.
- If this is AIX then ensure full core dumps are enabled - they are not enabled by default.
- A client-side timeout / disconnect.
- Unlikely bit possible: A fatal breakdown in SSL packet negotiation (though only if TLS is used)
It is also worth checking with "netstat" as well to see of you have processes stuck for any length of time in a TIME_WAIT state. On Windows systems the timeout interval can be adjusted in the registry and I set it very low by choice to < 15 seconds.
It is also worth reviewing the UniVerse 12 release notices on the
Product Availability Matrix (PAM) under
RESOURCES / PRODUCT AVAILABILITY and 'Available Hotfixes' as there are some fixes in 12.1.1.1018 that could be relevant.
Regards
JJ
------------------------------
John Jenkins
Thame, Oxfordshire
------------------------------
Matthew,
Difficult to tell in isolation and unless easily repeatable so it can be monitored, but the most likely candidates are all centred around a breakdown in the data exchange between client and server processes, including:
- The TCP connection between the client and server processes was abruptly terminated at the network level
- The client process unexpectedly terminated - if this is the case then there may be entries n the client system error log files.
- JAVA resource tuning or web server configuration can be factors here.
- The server process unexpectedly terminated - if this is the case then look at the server log files and for possible core files being generated at the time.
- If this is AIX then ensure full core dumps are enabled - they are not enabled by default.
- A client-side timeout / disconnect.
- Unlikely bit possible: A fatal breakdown in SSL packet negotiation (though only if TLS is used)
It is also worth checking with "netstat" as well to see of you have processes stuck for any length of time in a TIME_WAIT state. On Windows systems the timeout interval can be adjusted in the registry and I set it very low by choice to < 15 seconds.
It is also worth reviewing the UniVerse 12 release notices on the
Product Availability Matrix (PAM) under
RESOURCES / PRODUCT AVAILABILITY and 'Available Hotfixes' as there are some fixes in 12.1.1.1018 that could be relevant.
Regards
JJ
------------------------------
John Jenkins
Thame, Oxfordshire
------------------------------
Thanks for the detailed response John - we'll starting by looking at available patches.
------------------------------
Matthew Wright
Analyst
Tomago Aluminium Co PTY Ltd.
Tomago NSW AU
------------------------------
