This content has been marked as final. Show 5 replies
I also try to use plockstat on OS 2009.06 for debugging of my application and I also see exactly the same message like you:
Any help would be appreciated.
plockstat: processing aborted: Abort due to drop
Are either of you seeing any output prior to this message? Judging from the code, there should be some prior diagnostic information.
Certainly I see it w/o any additional output if I'm using -A parameter. If I lower my requirements to just -C it's running fine. Karel
I was asking for diagnostic output prior to the error message you're seeing. The reason is that in the code, there appears to be a message that should print before this one. Without that information, it's hard to see exactly why this error is printing.
This message is tied to a code macro EDT_DROPABORT. It seems it should occur only if there is a null error handler when one of the "drop" routines is called. It's not clear to me why this should occur, so I was hoping for more information from your output.
Without that piece of information, I only have a semi-informed, obvious guess. To wit, the plockstat request overruns the maximum buffer space allotted to DTrace. Rather than print drop information, plockstat appears to prefer aborting entirely. Since your less demanding call is working, that would seem to be the case.
You might try tuning the buffer size/policy to accommodate the request with the -x option. I don't know offhand which Solaris 10 updates have it.
as I said, plockstat prints just this line about abort alone. I think you are hitting nail on the head assuming this is due to dtrace buffers overrun, since if I insrease bufsize (and also aggsize or something like that -- found on inet) I don't get the abort that soon, but on the other hand I'm also not able to complete w/o abort. That's the reason I give up using -A and switched to use -C.