November 2014 - Asg - mailman3.cels.anl.gov

[aesop] #38: bracketed single-line if statements
by aesop 11 May '15

11 May '15

#38: bracketed single-line if statements --------------------------------+------------------------------------------- Reporter: gcdanie@… | Type: defect Status: new | Priority: don't know Component: don't know | Version: Keywords: | --------------------------------+------------------------------------------- When compiling a non-blocking function (e.g., without a __blocking identifier), should that function contain a single line if statement enclosed in brackets (as below), the aesop compiler will error (as below). if (condition) { return something; } aatree.c:79:1: error: stray ‘#’ in program aatree.c:79:7: warning: statement with no effect [-Wunused-value] aatree.c:79:10: error: expected ‘;’ before string constant aatree.c:81:31: error: ‘dir’ undeclared (first use in this function) aatree.c:81:31: note: each undeclared identifier is reported only once for each function it appears in aatree.c: In function ‘__ae_worker_treenode_traverse’: -- Ticket URL: <http://trac.mcs.anl.gov/projects/aesop/ticket/38> aesop <https://trac.mcs.anl.gov/projects/aesop> The Aesop Language

2 3

[aesop] #46: parallelfor2 test failure
by aesop 20 Mar '15

20 Mar '15

#46: parallelfor2 test failure -------------------------+-------------------------------------------------- Reporter: carns | Owner: Type: defect | Status: new Priority: minor | Component: don't know Version: devel branch | Keywords: -------------------------+-------------------------------------------------- The parser/tests/blocking/parallelfor2 test program fails with the following assertion: {{{ parallelfor2: /home/pcarns/working/aesop-devel/build/include/aesop/ae- ctl.h:229: ae_ctl_done: Assertion `refcount >= 0' failed. Aborted (core dumped) }}} -- Ticket URL: <https://trac.mcs.anl.gov/projects/aesop/ticket/46> aesop <https://trac.mcs.anl.gov/projects/aesop> The Aesop Language

1 3

Re: [ASG] Weekly Update
by Phil Carns 26 Nov '14

26 Nov '14

Thanks Jongmin. Is there really going to be a client side operation that will atomically create(a/b/c) if a and b don't yet exist? Or is this more of a mechanism for POSIX users to group arbitrary operations together into a transaction? thanks, -Phil On 11/25/2014 09:35 PM, Gim Jongmin wrote: > Hi Phil, > This is an example. A client sends create(a/b/c). However, the > directory 'a' and 'b' are not exist yet. In this situation, POSIX > tries to create the directory, it depends on create option, finally, > this create request implies mkdir 'a' , mkdir 'b', and create 'c' at > the same time. If we consider these operations individually, we can > create 3 logs for dir 'a', dir 'b', and file 'c'. If we met a failure > to write dir 'b', we are able to recover dir 'b' by the logging. > However, it may be failed for the request from the client application > since we could not recover file 'c'. So we want to make a transaction > for the request in the client FS layer. In the first place, we do not > send each write to server before finishing the transaction. The log > buffer holds it and write the log to server with just 1 write request. > For this, we have to put these 3 metadata on single array by > marshalling/unmarshalling and treat these as one I/O request. If it > was success, we are able to recover dir 'a', dir 'b', and file 'c'. If > the log writing was failed, it is same to 'undo'. The inconsistency > for the request may be solved by this suggestion. > We want to provide a transaction write for updating some related > files. I am writing a sample code now and also have a room for the > redesign. > Thanks, > Jongmin > > On Tue, Nov 25, 2014 at 7:45 PM, Phil Carns <carns(a)mcs.anl.gov > <mailto:[email protected]>> wrote: > > Hi Jongmin, > > Do you have some examples in mind of POSIX operations that will > use the metadata logging functionality? > > thanks, > -Phil > > > On 11/24/2014 06:25 PM, Gim Jongmin wrote: >> Dear All, >> >> Last week, we are trying to design a metadata logging to support >> a transaction, which is enable to support write completion for >> related multiple write, i..e, updating metadata with its parent >> metadata. We analyzed FUSE operation when read and write requests >> are arrived, and write code for a log pool. Also we are inserting >> a code to check whether the logging transaction is properly >> working or not. This work will be continued in this week. >> >> Sincerely, >> Jongmin >> >> -- >> Jongmin Gim >> Post-Doctoral Research Associate >> Dept. of Elecrical and Computer Enginerring >> Texas A&M University >> 333B Wisenbaker >> College Station, TX 77843-3259 >> Phone: 979-587-8311 <tel:979-587-8311> >> > > > _______________________________________________ > ASG mailing list > ASG(a)lists.mcs.anl.gov <mailto:[email protected]> > https://lists.mcs.anl.gov/mailman/listinfo/asg > > > > > -- > Jongmin Gim > Post-Doctoral Research Associate > Dept. of Elecrical and Computer Enginerring > Texas A&M University > 333B Wisenbaker > College Station, TX 77843-3259 > Phone: 979-587-8311 >

1 0

ANL Weekly Status - 20141124
by Harms, Kevin N. 26 Nov '14

26 Nov '14

Triton: * Metadata/Graph work presented at PDSW'14 * Informal meetup with Shane and Alex at SC * Planning Triton demo for some time in near future - considering "application" to test * continuing preliminary integration of SOS into Triton Simulation: * SWIM work presented at PMBS'14

1 0

Re: [ASG] Weekly Update
by Phil Carns 25 Nov '14

25 Nov '14

Hi Jongmin, Do you have some examples in mind of POSIX operations that will use the metadata logging functionality? thanks, -Phil On 11/24/2014 06:25 PM, Gim Jongmin wrote: > Dear All, > > Last week, we are trying to design a metadata logging to support a > transaction, which is enable to support write completion for related > multiple write, i..e, updating metadata with its parent metadata. We > analyzed FUSE operation when read and write requests are arrived, and > write code for a log pool. Also we are inserting a code to check > whether the logging transaction is properly working or not. This work > will be continued in this week. > > Sincerely, > Jongmin > > -- > Jongmin Gim > Post-Doctoral Research Associate > Dept. of Elecrical and Computer Enginerring > Texas A&M University > 333B Wisenbaker > College Station, TX 77843-3259 > Phone: 979-587-8311 >

2 1

Weekly Report
by Luogeng Dai 24 Nov '14

24 Nov '14

Dear All, Last week, I were still working on capability cache and authorization server prototyping. Regards, Luogeng

1 0

Weekly Update
by Gim Jongmin 24 Nov '14

24 Nov '14

Dear All, Last week, we are trying to design a metadata logging to support a transaction, which is enable to support write completion for related multiple write, i..e, updating metadata with its parent metadata. We analyzed FUSE operation when read and write requests are arrived, and write code for a log pool. Also we are inserting a code to check whether the logging transaction is properly working or not. This work will be continued in this week. Sincerely, Jongmin -- Jongmin Gim Post-Doctoral Research Associate Dept. of Elecrical and Computer Enginerring Texas A&M University 333B Wisenbaker College Station, TX 77843-3259 Phone: 979-587-8311

1 0

AU Update
by Shane Farmer 24 Nov '14

24 Nov '14

All, This week we are integrating learning from SC14, working on our new test suite to support new internal architectures, training our new resources, and toying with new possible examples and docs, as well as making small steps forward on some publishable work (more soon). Shane

1 0

Weekly report
by Guoxin Liu 24 Nov '14

24 Nov '14

Dear All, In the last week, we were continuing implementing the distributed simulator, our load balancing algorithm and the comparison algorithms to validate the effectiveness and efficiency of our design.. Best Regards! Guoxin Liu PHD Candidate ECE, Clemson University

1 0

20141124 - Sandia Status
by Danielson, Geoffrey Charles 24 Nov '14

24 Nov '14

All, Last week Sandia continued working on bringing back our client infrastructure and revising the transaction semantics in the storage server. Regards, Geoff Danielson gcdanie(a)sandia.gov Geoffrey.Danielson(a)hp.com Cell: 505-263-0751

1 0