There are 2 servers in the system and number of client computers. The
clients are booting via DHCPD from one of the servers.
The servers have 3 network cards connected to different networks (both
servers are on same networks, netmap is configured properly) and clients are
connected to one of the network.
There are some scripts running during intialization on the client, including
copying number of files from server to ramdisk on the client.
Time to time we face following problem. It comes in some “batches”, for
example we can repeat problem for 30 minutes and then it is gone.
Sometimes we can repeat it only once in number of retries.
If there is such problem, there is a process on the client REPLY blocked to
virtual circuit. For example “ksh” blocked to “Fsys” on the server.
If we do “sin vc” on the client, the virtual circuit is shown as REPLY/3.
The other end of the virtual circuit on the server is READY/1 and Fsys is
working normally.
The virtual circuit is blocked in REPLY/3 for lot of minutes.
If we do “netinfo” on the client, there are big number of same messages each
second:
(2325) 82557 (rx) data discarded - invalid vid buf This message is
reported from netinfo on the client, but from network cards on both servers
(MAC addresses), not from network card on the client.
Only solution is to restart the client computer. Very often the problem
appears again but not always for same process - there are more scripts and
utilities invoked during initialization of the client.
Few times we found another message on the server at same time:
NET (tx) failed (send) local process got eoi
2)
There are some background scripts running on the servers, performing some
operations in a loop. There is “sleep” invoked to block the loop for few
seconds.
Sometimes happens that sleep is not terminated. For example “sleep 5” was
running for more than few hours.
If we do “sin rt”, we can find that the “sleep” has timer attached and it is
expired:
//1/bin/sleep signal 14 -.— -.—
Command “sin sig” for the “sleep”:
SIG IGN SIG MASK SIG PEND SEG:OFF
00380000 00000000 00000000 000F:00CC8C
There are enough system resources available. Proc is running with -P2000
and -t512. Only cca 1000 proc entries and 260 timers were allocated.
3)
There is QNX 4.25 patch E running on the computers.
Network drivers on the servers: 2x Net.ether82557, 1x Net.epic.
Network driver on the clients: Net.ether82557
4)
Please send us as detailed as possible description for messages from the
“netinfo” utility - see the attached text file.
If you have description for all possible messages, it would be very helpfull
for us. We do not know what message is normal, warning or error.