Q_CAT:1447 Semaphore appears stuck

[cra]

Hello, does anybody know what might be the cause for the spurious error message "Q_CAT:1447 Semaphore appears stuck" issued by TMQFORWARD? We run Tuxedo 6.5 (yes, I know... with 10 TMQFORWARDs (polling @ 1 second interval) and each of them (sometimes also a TMS_QM) sometimes complains about a stuck semaphore. The "funny" thing is that this started suddenly at one day after months of operation without these warnings.

We tried reboots, complete rebuild of queue file, tlog file, tuxconfig file, nothing helped.

However, the system runs stable, there are no errors or timeouts. What can be done before contacting BEA support as suggested in the Docs? Changing IPC settings?

Thanks in advance for your help!

 

[piljak]

I am just administering a system that uses Tuxedo 6.4, have only a vague idea what it does.

We had the same problem and it also appeared all of a sudden and is not related to any changes on the system. And it did crash our application. We are running SCO OpenServer 5.0.6 .
BEA techs advised that we should change some kernel parameters (MSGMNB and MSGMNI) and add a SPINCOUNT 10000 setting to each machine's entry, also to change the -i parameter to have the value 5 for each TMQFORWARD process in configuration file.
I just came back from the test machine and none of the above helped. Error message still show up in ULOG.
Also from what I understand this was discovered first time on Sun Solaris.

I know it is not much help, but at least now you know you are not alone ;-).