*Apparent* Informix Network Daemon Failure

Once again we have encountered what is apparently an Informix Network Daemon failure. I say "apparently" because we have no error messages in the online.log or, as far as we can see, the UNIX system logs (we are still investigating), to prove this categorically. Users already connected to the server (via Telnet) mostly had their sessions disconnected, but we are still able to login via the UNIX console. Our main application connects via Shared Memory. Unfortunately we didn't think to "ping" the server from another server whilst it was in this state - we were more concerned about getting it back up again. The previous time this happened (last week), we got the following message on the online.log :- 13:04:08 listener-thread: err = -27100: oserr = 5: errstr = : InternalCommunications Error: NSF subsystem error. System error = 5. 13:04:08 Attempting to bring listener thread down. 13:04:08 Server stopped. Anybody got any ideas as to what may be causing this? Indeed, is there any command I can run to check whether the problem is Informix- generated or UNIX-generated. I feel sure this won't be the last such failure, as we have encountered 6 in the last 5 working days. TIA Mr Creosote SCO Openserver 5.0.4 Informix Dynamic Server 7.31.UC2 -- "Just a waffer thin mint?" Sent via Deja.com http://www.deja.com/ Before you buy.

In article <88tun3$q0m$1@nnrp1.deja.com>, Mister Creosote <glynb@my- deja.com> writes >Once again we have encountered what is apparently an Informix Network >Daemon failure. I say "apparently" because we have no error messages in >the online.log or, as far as we can see, the UNIX system logs (we are >still investigating), to prove this categorically. Users already >connected to the server (via Telnet) mostly had their sessions >disconnected, but we are still able to login via the UNIX console. Our >main application connects via Shared Memory. Unfortunately we didn't >think to "ping" the server from another server whilst it was in this >state - we were more concerned about getting it back up again. The >previous time this happened (last week), we got the following message >on the online.log :- > >13:04:08 listener-thread: err = -27100: oserr = 5: errstr = : Internal>Communications Error: NSF subsystem error. > System error = 5. >13:04:08 Attempting to bring listener thread down. > >13:04:08 Server stopped.> >Anybody got any ideas as to what may be causing this? Indeed, is there >any command I can run to check whether the problem is Informix- >generated or UNIX-generated. I feel sure this won't be the last such >failure, as we have encountered 6 in the last 5 working days. > >TIA > >Mr Creosote > >SCO Openserver 5.0.4 >Informix Dynamic Server 7.31.UC2 > Don't user 7.31.UC2 it contains a bug where doing a alter table, add a column and modify a decimal in the same statement will corrupt the table beyond the ability for Tech Support to patch it! 7.31.UC5 is available. Upgrade ASAP.. >-- >"Just a waffer thin mint?" > > >Sent via Deja.com http://www.deja.com/ >Before you buy. -- David Williams

Mister Creosote wrote: > > Once again we have encountered what is apparently an Informix Network > Daemon failure. I say "apparently" because we have no error messages in > the online.log or, as far as we can see, the UNIX system logs (we are > still investigating), to prove this categorically. Users already > connected to the server (via Telnet) mostly had their sessions > disconnected, but we are still able to login via the UNIX console. Our > main application connects via Shared Memory. Unfortunately we didn't > think to "ping" the server from another server whilst it was in this > state - we were more concerned about getting it back up again. The > previous time this happened (last week), we got the following message > on the online.log :- > > 13:04:08 listener-thread: err = -27100: oserr = 5: errstr = : Internal> Communications Error: NSF subsystem error. > System error = 5. > 13:04:08 Attempting to bring listener thread down. > > 13:04:08 Server stopped.> > Anybody got any ideas as to what may be causing this? Indeed, is there > any command I can run to check whether the problem is Informix- > generated or UNIX-generated. I feel sure this won't be the last such > failure, as we have encountered 6 in the last 5 working days. Do you have the Informix software and shared libraries installed on an NFS mounted volume? If yes, reinstall on a local drive. Any NFS error will cause the executable to fail when it tries to access its code or a shared library segment that has not already been loaded. Remember UNIX is demand paged! The entire executable is NOT loaded at start up time. -- Art S. Kagel & Family kagel@erols.com