Hello dtrace-discuss,
http://www-03.ibm.com/servers/aix/6/preview.html
"Dynamic tracing AIX 6 provides a new dynamic tracing capability that can
simplify
debugging complex system or application code. This dynamic tracing
facility will be introduced via a new tracing command, probevue,
that allows a developer or system administrator to dynamically insert
trace breakpoints in existing code without having to recompile the code."
dtrace like?
--
Best regards,
Robert mailto:rmilkowski at task.gda.pl
http://milek.blogspot.com
On Wed, May 23, 2007 at 09:52:34AM +0200, Robert Milkowski wrote:> Hello dtrace-discuss, > > http://www-03.ibm.com/servers/aix/6/preview.html > "Dynamic tracing AIX 6 provides a new dynamic tracing capability that can simplify > debugging complex system or application code. This dynamic tracing > facility will be introduced via a new tracing command, probevue, > that allows a developer or system administrator to dynamically insert > trace breakpoints in existing code without having to recompile the code." > > > dtrace like?I think it''s safe to say that IBM is ignoring my suggestion: http://blogs.sun.com/bmc/entry/dtrace_on_aix As for "probevue", we know very little about it (googling for "probevue" current yields three hits: one is the link you quoted, one is some blog entry saying that this "sure as hell sounds like DTrace" and the last is in Russian). From the little that we do know, it seems unlikely to be a holistic DTrace-like solution, but we''ll see... - Bryan -------------------------------------------------------------------------- Bryan Cantrill, Solaris Kernel Development. http://blogs.sun.com/bmc
Hello Bryan, Wednesday, May 23, 2007, 10:02:47 AM, you wrote: BC> On Wed, May 23, 2007 at 09:52:34AM +0200, Robert Milkowski wrote:>> Hello dtrace-discuss, >> >> http://www-03.ibm.com/servers/aix/6/preview.html >> "Dynamic tracing AIX 6 provides a new dynamic tracing capability that can simplify >> debugging complex system or application code. This dynamic tracing >> facility will be introduced via a new tracing command, probevue, >> that allows a developer or system administrator to dynamically insert >> trace breakpoints in existing code without having to recompile the code." >> >> >> dtrace like?BC> I think it''s safe to say that IBM is ignoring my suggestion: BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix BC> As for "probevue", we know very little about it (googling for "probevue" BC> current yields three hits: one is the link you quoted, one is some BC> blog entry saying that this "sure as hell sounds like DTrace" and the BC> last is in Russian). From the little that we do know, it seems unlikely BC> to be a holistic DTrace-like solution, but we''ll see... Perhaps it would be interesting if someone would install and check AIX 6 Open Beta - I don''t have a free POWER server nearby. -- Best regards, Robert mailto:rmilkowski at task.gda.pl http://milek.blogspot.com
On Wed 23 May 2007 at 10:33AM, Robert Milkowski wrote:> > Perhaps it would be interesting if someone would install and check AIX 6 Open > Beta - I don''t have a free POWER server nearby.According to the website, it will be available later in the year; I don''t think it is yet. -dp -- Daniel Price - Solaris Kernel Engineering - dp at eng.sun.com - blogs.sun.com/dp
After IBM introduces this probevue, dose it mean there exists a kernel-supported dynamic instrumentation tools in all Unix/Linux OS? Solairs, FreeBSD, MacOS - DTrace Linux - kprobe-based SystemTap AIX - probevue HPUX ... no more development on this system in hp, we will ignore this soon... 2007/5/23, Bryan Cantrill <bmc at eng.sun.com>:> > On Wed, May 23, 2007 at 09:52:34AM +0200, Robert Milkowski wrote: > > Hello dtrace-discuss, > > > > http://www-03.ibm.com/servers/aix/6/preview.html > > "Dynamic tracing AIX 6 provides a new dynamic tracing capability that > can simplify > > debugging complex system or application code. This dynamic tracing > > facility will be introduced via a new tracing command, probevue, > > that allows a developer or system administrator to dynamically insert > > trace breakpoints in existing code without having to recompile the > code." > > > > > > dtrace like? > > I think it''s safe to say that IBM is ignoring my suggestion: > > http://blogs.sun.com/bmc/entry/dtrace_on_aix > > As for "probevue", we know very little about it (googling for "probevue" > current yields three hits: one is the link you quoted, one is some > blog entry saying that this "sure as hell sounds like DTrace" and the > last is in Russian). From the little that we do know, it seems unlikely > to be a holistic DTrace-like solution, but we''ll see... > > - Bryan > > -------------------------------------------------------------------------- > Bryan Cantrill, Solaris Kernel Development. http://blogs.sun.com/bmc > _______________________________________________ > dtrace-discuss mailing list > dtrace-discuss at opensolaris.org >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20070523/fd42a387/attachment.html>
>After IBM introduces this probevue, dose it mean there exists a kernel-supported dynamic >instrumentation tools in all Unix/Linux OS?well, we will have to wait and see. I bet ''probevue'' will be hardwired to the P5,P6 and AIX kernel , question is: would IBM open source the project ? As already commented by Bryan, they seems to ignore to open source the AIX. stefan -- This message posted from opensolaris.org
I don''t think IBM has reasons to open source the AIX. And why does Sun open source Solaris, SPARC, Java in and after May. 2005? 2007/5/23, Stefan Parvu <stefanparvu14 at yahoo.com>:> > >After IBM introduces this probevue, dose it mean there exists a > kernel-supported dynamic >instrumentation tools in all Unix/Linux OS? > > well, we will have to wait and see. I bet ''probevue'' will be hardwired to > the P5,P6 and AIX kernel > , question is: would IBM open source the project ? As already commented by > Bryan, they seems > to ignore to open source the AIX. > > stefan > -- > This message posted from opensolaris.org > _______________________________________________ > dtrace-discuss mailing list > dtrace-discuss at opensolaris.org >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20070523/181535b1/attachment.html>
?? Euler Tao wrote:> And why does Sun open source Solaris, SPARC, Java in and after May. 2005?http://blogs.sun.com/jonathan/entry/what_we_did http://blogs.sun.com/jonathan/entry/censoring_free_media_or_fighting -- -Alan Coopersmith- alan.coopersmith at sun.com Sun Microsystems, Inc. - X Window System Engineering
I''ve read the articles. Thank you :) And now I know the reason that Sun opens source Solaris and the other hard/softwares. In a word, opensource makes Solaris attract more users and opensource makes Sun walk further than old years! It''s great. 2007/5/23, Alan Coopersmith <alan.coopersmith at sun.com>:> > ?? Euler Tao wrote: > > And why does Sun open source Solaris, SPARC, Java in and after May. > 2005? > > http://blogs.sun.com/jonathan/entry/what_we_did > http://blogs.sun.com/jonathan/entry/censoring_free_media_or_fighting > > -- > -Alan Coopersmith- alan.coopersmith at sun.com > Sun Microsystems, Inc. - X Window System Engineering >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20070523/af186c3a/attachment.html>
> BC> I think it''s safe to say that IBM is ignoring my suggestion: > > BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix > > BC> As for "probevue", we know very little about it (googling for "probevue" > BC> current yields three hits: one is the link you quoted, one is some > BC> blog entry saying that this "sure as hell sounds like DTrace" and the > BC> last is in Russian). From the little that we do know, it seems unlikely > BC> to be a holistic DTrace-like solution, but we''ll see... > > Perhaps it would be interesting if someone would install and check AIX 6 Open > Beta - I don''t have a free POWER server nearby.Yes, that would be great. I don''t think the Open Beta is available yet, but if anyone is both a DTrace user and an AIX customer, it would be great to see a comparison of probevue and DTrace once the former is in Beta... - Bryan -------------------------------------------------------------------------- Bryan Cantrill, Solaris Kernel Development. http://blogs.sun.com/bmc
On Wed, May 23, 2007 at 09:45:41AM -0700, Bryan Cantrill wrote:> > > BC> I think it''s safe to say that IBM is ignoring my suggestion: > > > > BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix > > > > BC> As for "probevue", we know very little about it (googling for "probevue" > > BC> current yields three hits: one is the link you quoted, one is some > > BC> blog entry saying that this "sure as hell sounds like DTrace" and the > > BC> last is in Russian). From the little that we do know, it seems unlikely > > BC> to be a holistic DTrace-like solution, but we''ll see... > > > > Perhaps it would be interesting if someone would install and check AIX 6 Open > > Beta - I don''t have a free POWER server nearby. > > Yes, that would be great. I don''t think the Open Beta is available yet, > but if anyone is both a DTrace user and an AIX customer, it would be great > to see a comparison of probevue and DTrace once the former is in Beta...Sounds interesting - if someone does check out probevue, the following are some of the Solaris DTrace features to compare against: Solaris DTrace features (mostly from memory), - dynamic kernel tracing (fbt:::) - static kernel tracing (sdt:::) - kernel profiling (profile:::) - high resolution profiling (cyclics) - anonymous tracing (dtrace -a) - postmortem tracing (mdb) - dynamic process tracing (pid:::) - static process tracing (sdt:::) - dynamic process instruction tracing (pid:::) - process profiling (profile:::) - syscall tracing (syscall:::) - kernel lock tracing (lockstat:::) - user lock tracing (plockstat:::) - system info tracing (sysinfo:::) - virtual memory tracing (vminfo:::) - scheduler tracing (sched:::) - process event tracing (proc:::) - I/O tracing (io::: disk/nfs) - filesystem tracing (fsinfo:::, fbt:::) - system event tracing (sysevent:::) - Java tracing (hotspot*:::) - JavaScript tracing (javascript*:::) - Ruby tracing (ruby*:::) - iSCSI tracing (iscsitgt*:::) - X11 tracing (Xserver*:::) - kernel stack tracing (stack()) - process stack tracing (ustack()) - java stack tracing (jstack()) - aggregations (@, printa()) - predicates (//) - thread local storage (self->) - clause local storage (this->) - speculative tracing (speculate()) - is-enabled probes (*_ENABLED) - switch buffers (bufpolicy=switch) - ring buffers (bufpolicy=ring) - fill buffers (bufpolicy=fill) - translators (translator <>) - privileges (dtrace_proc, ...) - DTrace test suite (SUNWdtrt) - scriptability (#!/usr/sbin/dtrace -s) - cacheable predicates - stability framework - JNI Binding - chime GUI A few of those features the latest Solaris bits, and/or additional providers to be downloaded and added. Sorry if I missed something important.. And of course, there are more DTrace features in the works. Perhaps worth mentioning are the indirect features one gets with DTrace, such as: - DTrace community (us!) - Dynamic Tracing Guide (PDF/html, docs.sun.com) - DTraceToolkit (100+ scripts) - Production since 2005 Brendan -- Brendan [CA, USA]
oh my god! half of these are not listed on the sun''s dtrace documentation. I''m now using the S10u3 :) 2007/5/24, Brendan Gregg - Sun Microsystems <brendan at sun.com>:> > On Wed, May 23, 2007 at 09:45:41AM -0700, Bryan Cantrill wrote: > > > > > BC> I think it''s safe to say that IBM is ignoring my suggestion: > > > > > > BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix > > > > > > BC> As for "probevue", we know very little about it (googling for > "probevue" > > > BC> current yields three hits: one is the link you quoted, one is > some > > > BC> blog entry saying that this "sure as hell sounds like DTrace" and > the > > > BC> last is in Russian). From the little that we do know, it seems > unlikely > > > BC> to be a holistic DTrace-like solution, but we''ll see... > > > > > > Perhaps it would be interesting if someone would install and check AIX > 6 Open > > > Beta - I don''t have a free POWER server nearby. > > > > Yes, that would be great. I don''t think the Open Beta is available yet, > > but if anyone is both a DTrace user and an AIX customer, it would be > great > > to see a comparison of probevue and DTrace once the former is in Beta... > > Sounds interesting - if someone does check out probevue, the following are > some of the Solaris DTrace features to compare against: > > Solaris DTrace features (mostly from memory), > > - dynamic kernel tracing (fbt:::) > - static kernel tracing (sdt:::) > - kernel profiling (profile:::) > - high resolution profiling (cyclics) > - anonymous tracing (dtrace -a) > - postmortem tracing (mdb) > - dynamic process tracing (pid:::) > - static process tracing (sdt:::) > - dynamic process instruction tracing (pid:::) > - process profiling (profile:::) > - syscall tracing (syscall:::) > - kernel lock tracing (lockstat:::) > - user lock tracing (plockstat:::) > - system info tracing (sysinfo:::) > - virtual memory tracing (vminfo:::) > - scheduler tracing (sched:::) > - process event tracing (proc:::) > - I/O tracing (io::: disk/nfs) > - filesystem tracing (fsinfo:::, fbt:::) > - system event tracing (sysevent:::) > - Java tracing (hotspot*:::) > - JavaScript tracing (javascript*:::) > - Ruby tracing (ruby*:::) > - iSCSI tracing (iscsitgt*:::) > - X11 tracing (Xserver*:::) > - kernel stack tracing (stack()) > - process stack tracing (ustack()) > - java stack tracing (jstack()) > - aggregations (@, printa()) > - predicates (//) > - thread local storage (self->) > - clause local storage (this->) > - speculative tracing (speculate()) > - is-enabled probes (*_ENABLED) > - switch buffers (bufpolicy=switch) > - ring buffers (bufpolicy=ring) > - fill buffers (bufpolicy=fill) > - translators (translator <>) > - privileges (dtrace_proc, ...) > - DTrace test suite (SUNWdtrt) > - scriptability (#!/usr/sbin/dtrace -s) > - cacheable predicates > - stability framework > - JNI Binding > - chime GUI > > A few of those features the latest Solaris bits, and/or additional > providers to be downloaded and added. > > Sorry if I missed something important.. And of course, there are more > DTrace features in the works. > > Perhaps worth mentioning are the indirect features one gets with DTrace, > such as: > > - DTrace community (us!) > - Dynamic Tracing Guide (PDF/html, docs.sun.com) > - DTraceToolkit (100+ scripts) > - Production since 2005 > > Brendan > > -- > Brendan > [CA, USA] > _______________________________________________ > dtrace-discuss mailing list > dtrace-discuss at opensolaris.org >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20070524/d6ab7caa/attachment.html>
oh!!! I''m wrong! some of this list are the features, not new providers. I''m careless just now... 2007/5/24, ?? Euler Tao <eulertao at gmail.com>:> > oh my god! half of these are not listed on the sun''s dtrace documentation. > I''m now using the S10u3 :) > > > 2007/5/24, Brendan Gregg - Sun Microsystems <brendan at sun.com>: > > > > On Wed, May 23, 2007 at 09:45:41AM -0700, Bryan Cantrill wrote: > > > > > > > BC> I think it''s safe to say that IBM is ignoring my suggestion: > > > > > > > > BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix > > > > > > > > BC> As for "probevue", we know very little about it (googling for > > "probevue" > > > > BC> current yields three hits: one is the link you quoted, one is > > some > > > > BC> blog entry saying that this "sure as hell sounds like DTrace" > > and the > > > > BC> last is in Russian). From the little that we do know, it seems > > unlikely > > > > BC> to be a holistic DTrace-like solution, but we''ll see... > > > > > > > > Perhaps it would be interesting if someone would install and check > > AIX 6 Open > > > > Beta - I don''t have a free POWER server nearby. > > > > > > Yes, that would be great. I don''t think the Open Beta is available > > yet, > > > but if anyone is both a DTrace user and an AIX customer, it would be > > great > > > to see a comparison of probevue and DTrace once the former is in > > Beta... > > > > Sounds interesting - if someone does check out probevue, the following > > are > > some of the Solaris DTrace features to compare against: > > > > Solaris DTrace features (mostly from memory), > > > > - dynamic kernel tracing (fbt:::) > > - static kernel tracing (sdt:::) > > - kernel profiling (profile:::) > > - high resolution profiling (cyclics) > > - anonymous tracing (dtrace -a) > > - postmortem tracing (mdb) > > - dynamic process tracing (pid:::) > > - static process tracing (sdt:::) > > - dynamic process instruction tracing (pid:::) > > - process profiling (profile:::) > > - syscall tracing (syscall:::) > > - kernel lock tracing (lockstat:::) > > - user lock tracing (plockstat:::) > > - system info tracing (sysinfo:::) > > - virtual memory tracing (vminfo:::) > > - scheduler tracing (sched:::) > > - process event tracing (proc:::) > > - I/O tracing (io::: disk/nfs) > > - filesystem tracing (fsinfo:::, fbt:::) > > - system event tracing (sysevent:::) > > - Java tracing (hotspot*:::) > > - JavaScript tracing (javascript*:::) > > - Ruby tracing (ruby*:::) > > - iSCSI tracing (iscsitgt*:::) > > - X11 tracing (Xserver*:::) > > - kernel stack tracing (stack()) > > - process stack tracing (ustack()) > > - java stack tracing (jstack()) > > - aggregations (@, printa()) > > - predicates (//) > > - thread local storage (self->) > > - clause local storage (this->) > > - speculative tracing (speculate()) > > - is-enabled probes (*_ENABLED) > > - switch buffers (bufpolicy=switch) > > - ring buffers (bufpolicy=ring) > > - fill buffers (bufpolicy=fill) > > - translators (translator <>) > > - privileges (dtrace_proc, ...) > > - DTrace test suite (SUNWdtrt) > > - scriptability (#!/usr/sbin/dtrace -s) > > - cacheable predicates > > - stability framework > > - JNI Binding > > - chime GUI > > > > A few of those features the latest Solaris bits, and/or additional > > providers to be downloaded and added. > > > > Sorry if I missed something important.. And of course, there are more > > DTrace features in the works. > > > > Perhaps worth mentioning are the indirect features one gets with DTrace, > > such as: > > > > - DTrace community (us!) > > - Dynamic Tracing Guide (PDF/html, docs.sun.com) > > - DTraceToolkit (100+ scripts) > > - Production since 2005 > > > > Brendan > > > > -- > > Brendan > > [CA, USA] > > _______________________________________________ > > dtrace-discuss mailing list > > dtrace-discuss at opensolaris.org > > > >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20070524/b13c0c38/attachment.html>
> I think it''s safe to say that IBM is ignoring my suggestion: > > http://blogs.sun.com/bmc/entry/dtrace_on_aix > > As for "probevue", we know very little about it (googling for "probevue" > current yields three hits: one is the link you quoted, one is some > blog entry saying that this "sure as hell sounds like DTrace" and the > last is in Russian). From the little that we do know, it seems unlikely > to be a holistic DTrace-like solution, but we''ll see...http://www.linux.org.ru/view-message.jsp?msgid=1935398 The Russian page translation: --------------------------------------------------- "IBM tries to ''sort of open'' AIX 6 - the declaration of intentions. AIX 6 open beta will be available for the participants of the AIX 6 open beta program in mid 2007 (the http://www-03.ibm.com/servers/aix/ page will contain links to ISO images). It would be available for Power4, PPC970, POWER5 or POWER6 processors including the IBM System p, IBM System p5, IBM eServer p5, IBM eServer pSeries server product lines, as well as IBM BladeCenter JS2x blades and IntelliStation POWER workstations. What will be missing 1) Some crypto options (related to export control restrictions). Missing crypto options include Encrypted Filesystem, trusted Execution and NFS encryption support. 2) Flexible installation support. A simplified installation based on mksysb will be provided, everything will be simplified to the Vista level: - No "Secure by Default" and "Trusted AIX" templates - No installation GUI - The installer will try to remove everything on the target disk before installing No support for data migration when upgrading beta to the release AIX 6. <I am not sure whether the author means support from the system or support from IBM>. What will be included: 1) Itself 2) Workload partitions 3)Role Based Access Control 4)AIX Security Expert enhancements 5)Name Resolver Caching Daemon 6)probevue dynamic tracing 7)System Director Console for AIX P.S. IMHO this news should be treated not as a news about ISO images availability but as an attempt to increase interest towards AIX. ------------------------------------------------------------------------ There are tons of comments that are left as an exercise to the reader :-). One comment adds that the missing part includes full source code of AIX6 and most of the DDK. -- akolb
Thanks for that list.. mind if I use it in class? On 24/05/2007, at 1:47 PM, Brendan Gregg - Sun Microsystems wrote:> On Wed, May 23, 2007 at 09:45:41AM -0700, Bryan Cantrill wrote: >> >>> BC> I think it''s safe to say that IBM is ignoring my suggestion: >>> >>> BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix >>> >>> BC> As for "probevue", we know very little about it (googling for >>> "probevue" >>> BC> current yields three hits: one is the link you quoted, one >>> is some >>> BC> blog entry saying that this "sure as hell sounds like DTrace" >>> and the >>> BC> last is in Russian). From the little that we do know, it >>> seems unlikely >>> BC> to be a holistic DTrace-like solution, but we''ll see... >>> >>> Perhaps it would be interesting if someone would install and >>> check AIX 6 Open >>> Beta - I don''t have a free POWER server nearby. >> >> Yes, that would be great. I don''t think the Open Beta is >> available yet, >> but if anyone is both a DTrace user and an AIX customer, it would >> be great >> to see a comparison of probevue and DTrace once the former is in >> Beta... > > Sounds interesting - if someone does check out probevue, the > following are > some of the Solaris DTrace features to compare against: > > Solaris DTrace features (mostly from memory), > > - dynamic kernel tracing (fbt:::) > - static kernel tracing (sdt:::) > - kernel profiling (profile:::) > - high resolution profiling (cyclics) > - anonymous tracing (dtrace -a) > - postmortem tracing (mdb) > - dynamic process tracing (pid:::) > - static process tracing (sdt:::) > - dynamic process instruction tracing (pid:::) > - process profiling (profile:::) > - syscall tracing (syscall:::) > - kernel lock tracing (lockstat:::) > - user lock tracing (plockstat:::) > - system info tracing (sysinfo:::) > - virtual memory tracing (vminfo:::) > - scheduler tracing (sched:::) > - process event tracing (proc:::) > - I/O tracing (io::: disk/nfs) > - filesystem tracing (fsinfo:::, fbt:::) > - system event tracing (sysevent:::) > - Java tracing (hotspot*:::) > - JavaScript tracing (javascript*:::) > - Ruby tracing (ruby*:::) > - iSCSI tracing (iscsitgt*:::) > - X11 tracing (Xserver*:::) > - kernel stack tracing (stack()) > - process stack tracing (ustack()) > - java stack tracing (jstack()) > - aggregations (@, printa()) > - predicates (//) > - thread local storage (self->) > - clause local storage (this->) > - speculative tracing (speculate()) > - is-enabled probes (*_ENABLED) > - switch buffers (bufpolicy=switch) > - ring buffers (bufpolicy=ring) > - fill buffers (bufpolicy=fill) > - translators (translator <>) > - privileges (dtrace_proc, ...) > - DTrace test suite (SUNWdtrt) > - scriptability (#!/usr/sbin/dtrace -s) > - cacheable predicates > - stability framework > - JNI Binding > - chime GUI > > A few of those features the latest Solaris bits, and/or additional > providers to be downloaded and added. > > Sorry if I missed something important.. And of course, there are more > DTrace features in the works. > > Perhaps worth mentioning are the indirect features one gets with > DTrace, > such as: > > - DTrace community (us!) > - Dynamic Tracing Guide (PDF/html, docs.sun.com) > - DTraceToolkit (100+ scripts) > - Production since 2005 > > Brendan > > -- > Brendan > [CA, USA] > _______________________________________________ > dtrace-discuss mailing list > dtrace-discuss at opensolaris.org >
G''Day Boyd, On Mon, May 28, 2007 at 09:09:51PM +1000, Boyd Adamson wrote:> Thanks for that list.. mind if I use it in class?It''s not an official list, but if it helps - please do. I''m adding screenshots to each of the features to make the list more concrete: see http://www.solarisinternals.com/wiki/index.php/DTrace_Topics_Features. Brendan> On 24/05/2007, at 1:47 PM, Brendan Gregg - Sun Microsystems wrote:[...]> >Solaris DTrace features (mostly from memory), > > > > - dynamic kernel tracing (fbt:::) > > - static kernel tracing (sdt:::) > > - kernel profiling (profile:::) > > - high resolution profiling (cyclics) > > - anonymous tracing (dtrace -a) > > - postmortem tracing (mdb) > > - dynamic process tracing (pid:::) > > - static process tracing (sdt:::) > > - dynamic process instruction tracing (pid:::) > > - process profiling (profile:::)[...] -- Brendan [CA, USA]
Alexander Kolbasov wrote:> .... > >One comment adds that the missing part includes full source code of AIX6 and >most of the DDK. > >Interesting but that (nor any of the other bullet items) seem to address it being perhaps the worst platforms I''ve ever had to try and do kernel development on. Developer friendly? Not! Darren
> It''s not an official list, but if it helps - please > do. I''m adding > screenshots to each of the features to make the list > more concrete: see > http://www.solarisinternals.com/wiki/index.php/DTrace_ > Topics_Features. > > Brendan > > > On 24/05/2007, at 1:47 PM, Brendan Gregg - Sun > Microsystems wrote: > [...] > > >Solaris DTrace features (mostly from memory), > > > > > > - dynamic kernel tracing (fbt:::) > > > - static kernel tracing (sdt:::) > > > - kernel profiling (profile:::)HI Brendan, i may have missed it ? one of the nicest features i like about dtrace is that the variables, actual arguments in function calls, and kernel symbols are CTF empowered. also, i noticed from the various posts that the blurb from IBM mentioned dynamic insertion of probes into application code; not sure if that translates to insertions into live kernels as well. cheers, sam -- This message posted from opensolaris.org
something special,maybe this web will give you something useful:http://www.cntradecity.com/product.asp?Classid=4&Nclassid=110 -- This message posted from opensolaris.org
> > BC> I think it''s safe to say that IBM is ignoring my suggestion: > > > > BC> http://blogs.sun.com/bmc/entry/dtrace_on_aix > > > > BC> As for "probevue", we know very little about it (googling for "probevue" > > BC> current yields three hits: one is the link you quoted, one is some > > BC> blog entry saying that this "sure as hell sounds like DTrace" and the > > BC> last is in Russian). From the little that we do know, it seems unlikely > > BC> to be a holistic DTrace-like solution, but we''ll see... > > > > Perhaps it would be interesting if someone would install and check AIX 6 Open > > Beta - I don''t have a free POWER server nearby. > > Yes, that would be great. I don''t think the Open Beta is available yet, > but if anyone is both a DTrace user and an AIX customer, it would be great > to see a comparison of probevue and DTrace once the former is in Beta...IBM''s ProbeVue documentation is now public: http://publib.boulder.ibm.com/infocenter/pseries/v6r1/index.jsp?topic=/com.ibm.aix.genprogc/doc/genprogc/dynamic_tracing.htm&tocNode=int_9140 While it''s flattering to see that they are, um, "inspired" by DTrace, I stick to my original recommendation that they simply port DTrace to AIX -- it would be a lot less heartache, and instead of some crazy alien knockoff ("@@BEGIN"? "__k:"?!) they would have something that has a large, growing and (increasingly) cross-platform community. But I''m sure that ProbeVue and its accompanying VUE language(*) will be an improvement for AIX users regardless, so I guess they should just be relieved that IBM didn''t port TNF to AIX... - Bryan -------------------------------------------------------------------------- Bryan Cantrill, Solaris Kernel Development. http://blogs.sun.com/bmc (*) It''s unclear if "VUE" is an acronym for something or if someone''s caps-lock key was stuck and their spell checker was busted...
> IBM''s ProbeVue documentation is now public: > > http://publib.boulder.ibm.com/infocenter/pseries/v6r1/index.jsp?topic=/com.ibm.aix.genprogc/doc/genprogc/dynamic_tracing.htm&tocNode=int_9140> While it''s flattering to see that they are, um, "inspired" by DTrace, I > stick to my original recommendation that they simply port DTrace to AIX -- > it would be a lot less heartache, and instead of some crazy alien knockoff > ("@@BEGIN"? "__k:"?!) they would have something that has a large, growing > and (increasingly) cross-platform community. But I''m sure that ProbeVue > and its accompanying VUE language(*) will be an improvement for AIX users > regardless, so I guess they should just be relieved that IBM didn''t port > TNF to AIX...Interesting - they were not "inspired" by SDT/USDT probes - is it so boring? (Or they do support USDT?) Since AIX source is not available, how are users supposed to know which kernel functions to look at and what their arguments are? For the rest it seems like a simple sed(1) script would easily convert between VUE and D. - akolb