Hi Mike,
The default limit is 250,000 tracepoints (that is, points of instrumentation,
not probes). You can increase this by editing /kernel/drv/fasttrap.conf and
rebooting (there are other ways to cause the file to be reloaded, but
rebooting is the easiest and least prone to failure).
8588 sounds like quite a large number of USDT probes. While that certainly
should work, bear in mind that the pid provider can be used to trace any
function and any instruction; USDT should mostly be used to aggregate
points of semantic relevance. What is this provider exactly?
Adam
On Mon, Apr 14, 2008 at 08:42:27AM -0700, Mike Bell
wrote:> Is there a limit on the number of user space probes that can be present at
any time?
>
> I have compiled very many probes into our product, and can''t see
all of them in "dtrace -l". it looks very much like I have all of them
up to some point at which I''ve hit some hard limit.
>
> I can see a total of 8588 probes, I think 2 processes have all of the ones
I expect (one has 1752 probes, the other 2142), but my 3rd and largest process
only has 4694 probes. I haven''t checked but I expect it should have
somewhere between 10000 and 20000.
>
> Looking at some different systems, it seems the limit is on a single
process having more than 4694 probes. We are running Solaris 10 Update 3 I
believe, uname -a says:
> SunOS xxxxx 5.10 Generic_125101-10 i86pc i386 i86pc
>
> Is there any way to work around this limit, or do I need to be a bit less
ambitious with my probing?
>
> Thanks,
> Mike
>
>
> --
> This message posted from opensolaris.org
> _______________________________________________
> dtrace-discuss mailing list
> dtrace-discuss at opensolaris.org
--
Adam Leventhal, Fishworks http://blogs.sun.com/ahl