UBEs Fail on "msgmni" message

UBEs Fail on \"msgmni\" message

Our Enterprise Server is Sun UNIX. We are seeing an issue that occurs about twice a week. The server stops processing UBEs. After checking the UBE log, we see the following message:

17678 Tue Mar 5 11:03:46 2002 ipcsysv.c639
IPC2100017 createIPCMsgq (name Krnl17678RspQ) failed, errno=28: No space left on device.

17678 Tue Mar 5 11:03:46 2002 ipcsysv.c641
IPC2100034 Check OS kernel parameter msgmni.

JDE recommends bumping up the "msgmni" setting on the server. Which our server's setting is set to default(50). We are going to make the change this weekend using a formula that JDE sent us.

My question is, has anyone experienced this problem? Did bumping up the "msgmni" setting fix the problem?

Tim
XE SP16.1/Update1
SUN(UNIX)
Oracle 8.1.6
WTS(2000)

Re: UBEs Fail on \"msgmni\" message

Tim,

we have not experienced that nor are we a Sun shop. However, a stab in the dark idea based on the error message makes me ask you, what is the state of the system's virtual memory around 11:00am?

Larry Jones
[email protected]
OneWorld XE, SP 15.1
HPUX 11, Oracle SE 8.1.6
Mfg, Distribution, Financials

Re: UBEs Fail on \"msgmni\" message

I have seen upping this setting (and a few other kernel params)
be to low by default several times at different sites. In all
cases, increases the resource directed fixed the problem.
Generally we increased the params slightly more than recommended,
simply because IT at the site didn't always know what
IPC requirements their other vendor-based software might have.