Sun / Oracle problem

[nontrad]

I am getting a Sun adm message file with the following entries over and over. I have included one whole "cycle" of messages below.

What causes this? How can I fix the problem? (this message posted on an Oracle board also)

Thanks.
****************
Feb 14 12:06:02 oracle SSP Startup: : [ID 702911 local0.info] with SUNW_HOSTNAME = oracle
Feb 14 12:06:02 oracle SSP Startup: : [ID 702911 local0.info] REstarting: codd
Feb 14 12:06:03 oracle syslog: [ID 702911 local0.error] codd [oracle]: ERR: codd.c, 537: 1024:Can't get platform hostid; error: Could not find primary hostid from eeprom file
Feb 14 12:06:05 oracle syslog: [ID 702911 local0.warning] edd: WARNING: EDDApp.cc, 535: Waiting for cbs<->cbe connection...
Feb 14 12:06:10 oracle last message repeated 1 time
Feb 14 12:06:11 oracle cbs: [ID 702911 local0.error] cbs: ERR:[oracle]:cbs_get_cbe_ports:gethostbyname failed: host=oraclecb0 :Resource temporarily unavailable
Feb 14 12:06:11 oracle cbs: [ID 702911 local0.error] cbs: ERR:[oracle]:Cbmp_Connect:gethostbyname failed: hostname=oraclecb0
Feb 14 12:06:15 oracle syslog: [ID 702911 local0.warning] edd: WARNING: EDDApp.cc, 535: Waiting for cbs<->cbe connection...
Feb 14 12:06:20 oracle last message repeated 1 time
Feb 14 12:06:21 oracle cbs: [ID 702911 local0.error] cbs: ERR:[oracle]:cbs_get_cbe_ports:gethostbyname failed: host=oraclecb0 :Resource temporarily unavailable
Feb 14 12:06:21 oracle cbs: [ID 702911 local0.error] cbs: ERR:[oracle]:Cbmp_Connect:gethostbyname failed: hostname=oraclecb0
Feb 14 12:06:25 oracle syslog: [ID 702911 local0.warning] edd: WARNING: EDDApp.cc, 535: Waiting for cbs<->cbe connection...
Feb 14 12:06:30 oracle last message repeated 1 time
Feb 14 12:06:31 oracle cbs: [ID 702911 local0.error] cbs: ERR:[oracle]:cbs_get_cbe_ports:gethostbyname failed: host=oraclecb0 :Resource temporarily unavailable
Feb 14 12:06:31 oracle cbs: [ID 702911 local0.error] cbs: ERR:[oracle]:Cbmp_Connect:gethostbyname failed: hostname=oraclecb0

 

[MikeLacey]

Looks like a solaris problem rather than oracle.

I would check my TCP/IP setup -- can you ping localhost and your own server name for instance? What name does hostname (or uname) return? Is it the one you expect? Mike
michael.j.lacey@ntlworld.com
Email welcome if you're in a hurry or something -- but post in tek-tips as well please, and I will post my reply here as well.

 

[ChrisBroadway]

I'm confused .. unless I'm sadly mistaken this is the output from an SSP that would be used to control an E10k .. yet you mention that you've posted this on an Oracle board .. Why would you be running Oracle on an SSP ????
Can we clarify if this is an SSP ???

If this is definitely not an SSP try
pkginfo | grep ssp
anything that says System Service Processor is required
to look after an E10k .. Also look for (i think) SUNWcod which also looks to be installed .. I don't think that should be on anything over than an ssp yet.

Hope that helps

Chris

 

[nontrad]

After MUCH investigation. I found out that this IS an SSP. This office refers to this server as &quot;the Oracle server&quot; because it is dedicated to the purpose of running the 3 DBs only.I guess their assumption is that the (P/T student - me) Oracle DBA should know all about this stuff and handle not only the DBs but the OS &amp; network too.

The Oracle software is &quot;implemented&quot; via a telnet to the server that is running the system under a Unix OS.When I posted the message, I could not tell if this was something Oracle was telling me it needed or what.

As best as I can gather so far, when the server/Oracle was installed last summer, the EEPROM that came with it was not programmed with the server name via SSP and that is what is generating these messages every 30 seconds. The messages had been accumulating all this time and no one noticed until the file size overwhelmed the disk space.

E.

 

[ChrisBroadway]

Really !!??

If it is an SSP its got a weird naming convention .. why's everything called oracle

I would deduce that this message came from the SSP because 'cbs_get_cbe_port' is the control board server(running on SSP end) trying to talk to the control
board executive(running on the Platform).. But why all the comms failures .. Are you actually talking to the platform

can you ping the control boards (thats oraclecb0 and, I assume oraclecb1)?

If you su - ssp does domain_status, or fan, or hostinfo return a response...

If it is working ok as an SSP then the excessive errors may be down to COD or SSP patch levels.. Check with Sun for the latest and greatest

 

[MikeLacey]

What's an SSP then? (I'm an Oracle person really) Mike
michael.j.lacey@ntlworld.com
Email welcome if you're in a hurry or something -- but post in tek-tips as well please, and I will post my reply here as well.

 

[ChrisBroadway]

aaarrrrgghh a DBA on a sysadm board .. kill the heretic ;-)

Its really quite dull I'm afraid .. SSP is a System Service Processor.. basically the E10000 (Suns big server) consists of up to 16 system boards which can be logically partitioned into a number of domains (each domain being one or more system boards), with each domain appearing to be a server in its own right.. This means that your eeprom etc etc isn't stored on the E10k itself (coz we don't know how many domains there might be) so we have another sparc machine (usually an E250) called an SSP which allows us to monitor the whole E10K, create domains, boot domains etc etc

there we go .. something to impress the ladies with !

 

[MikeLacey]

Ahh... Hardware -- I don't do harware........

Thanks for the explanation Chris.

Regards,
Mike
michael.j.lacey@ntlworld.com
Email welcome if you're in a hurry or something -- but post in tek-tips as well please, and I will post my reply here as well.