From hayashemtov at yahoo.com Thu Oct 5 10:37:49 2006 From: hayashemtov at yahoo.com (haya shem tov) Date: Thu, 5 Oct 2006 07:37:49 -0700 (PDT) Subject: [Disksim-users] disk block address space Message-ID: <20061005143749.61873.qmail@web42408.mail.scd.yahoo.com> hi, I'm using disksim external interface with the st41601n.parv disk, which has a storage capacity of 2,676,846 blocks. when I'm addressing block 1,750,248 I got the following error: *** assertion failed: in disk_buffer_seekdone() (disksim_diskctlr.c:3925): ((first <= curr->blkno) && (curr->blkno <= last )): Mapping problem, block (1750248) not within track (1750249, 1750322) what went wrong? is that a malfunction? more generally, am I allowed to address each block within the Block count (2676846), and read, for example, 4096 bytes (one block)? thanks, haya --------------------------------- Want to be your own boss? Learn how on Yahoo! Small Business. -------------- next part -------------- An HTML attachment was scrubbed... URL: From bucy at gloop.org Thu Oct 5 14:38:39 2006 From: bucy at gloop.org (John Bucy) Date: Thu, 5 Oct 2006 11:38:39 -0700 Subject: [Disksim-users] disk block address space In-Reply-To: <20061005143749.61873.qmail@web42408.mail.scd.yahoo.com> References: <20061005143749.61873.qmail@web42408.mail.scd.yahoo.com> Message-ID: <4fc1e0430610051138i1890833cncbf0701d6932d4f0@mail.gmail.com> It sounds like that's a bug, perhaps in the layout code. In general, that model is *very* old. I think about the newest one in the 3.0 release distribution is the atlas10k model which you might want to try instead. And in response to your question, you definitely should be able to access every lbn from 0 to n-1. john On 10/5/06, haya shem tov wrote: > hi, > I'm using disksim external interface with the > st41601n.parv disk, which has a storage capacity of 2,676,846 blocks. > when I'm addressing block 1,750,248 I got the > following error: > *** assertion failed: in disk_buffer_seekdone() > (disksim_diskctlr.c:3925): ((first <= curr->blkno) && > (curr->blkno <= last )): > Mapping problem, block (1750248) not within track > (1750249, 1750322) > what went wrong? is that a malfunction? > more generally, am I allowed to address each block within > the Block count (2676846), > and read, for example, 4096 bytes (one block)? > thanks, > haya > > > ________________________________ > Want to be your own boss? Learn how on Yahoo! Small Business. > > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > From bucy at ece.cmu.edu Thu Oct 5 19:15:20 2006 From: bucy at ece.cmu.edu (John Bucy) Date: Thu, 5 Oct 2006 16:15:20 -0700 Subject: [Disksim-users] logical partitions In-Reply-To: <20060919234210.89700.qmail@web58507.mail.re3.yahoo.com> References: <20060919234210.89700.qmail@web58507.mail.re3.yahoo.com> Message-ID: <4fc1e0430610051615j110e0069sdf4f499300bedd00@mail.gmail.com> A simple partitioning scheme will split the underlying block device (disk, raid, etc) into contiguous ranges of LBNs. Disksim is just a block device so it doesn't know/care -- your requests to the first partition would be constrained to the first lbn range, etc. john On 9/19/06, micael yaborovsky wrote: > Disksim users, > I am new using disksim. > Spouse I would like to simulate logical partitions in the disk, > And two proceses: one sends requests to the first partition, and the other > to the second partition, how would I simulate this? > are logical partitions transparent to disksim? > Good day, > Micael yaborobsky > > > ________________________________ > Want to be your own boss? Learn how on Yahoo! Small Business. > > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > From bucy at ece.cmu.edu Thu Oct 5 19:18:41 2006 From: bucy at ece.cmu.edu (John Bucy) Date: Thu, 5 Oct 2006 16:18:41 -0700 Subject: [Disksim-users] synthetic workload In-Reply-To: <20060921142928.62975.qmail@web34002.mail.mud.yahoo.com> References: <20060921142928.62975.qmail@web34002.mail.mud.yahoo.com> Message-ID: <4fc1e0430610051618y2b4301d1r4258a124cbe951e6@mail.gmail.com> Because of the way the workload generator is integrated into the main disksim event loop, I expect it wouldn't be completely trivial to pull it out -- you'd probably have to reimplement some of the environment it expects to run in. But you can probably lift out code that computes the sizes/lbns/interarrival times/etc and wrap your own "driver harness" around it. john On 9/21/06, shay raz wrote: > Hello > I would like to input syssim, the disksim interface, with synthetic > workload. > is it possible to extract the synthetic workload code from disksim source? > what are the files involved, is it just disksim_synthio ? > does anybody have program for generating synthetic workload? > best, > Shay Raz > > ________________________________ > Want to be your own boss? Learn how on Yahoo! Small Business. > > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > From kamashay at yahoo.com Tue Oct 10 19:36:31 2006 From: kamashay at yahoo.com (shay raz) Date: Tue, 10 Oct 2006 16:36:31 -0700 (PDT) Subject: [Disksim-users] Disksim migration Message-ID: <20061010233631.66429.qmail@web34012.mail.mud.yahoo.com> Disksim users, I would like to migrate my disksim from old red-hat 7.3 to a newer platform like ubuntu. I installed on my ubuntu box gcc-2.95 and other tools like flex, bison etc. Yet, I got the following error, trying to ?make? disksim: lex.libparam.o lex.libparam.c libparam.lex: In function `libparamlex': libparam.lex:141: `yy_current_buffer' undeclared (first use in this function) libparam.lex:141: (Each undeclared identifier is reported only once libparam.lex:141: for each function it appears in.) libparam.lex: At top level: lex.libparam.c:1315: warning: `yyunput' defined but not used make[1]: *** [lex.libparam.o] Error 1 is that a version problem or some tool is missing? Is error in libparam.lex related to flex package? What should I do? Very best, shay --------------------------------- Want to be your own boss? Learn how on Yahoo! Small Business. -------------- next part -------------- An HTML attachment was scrubbed... URL: From bucy at gloop.org Tue Oct 10 19:42:01 2006 From: bucy at gloop.org (John Bucy) Date: Tue, 10 Oct 2006 16:42:01 -0700 Subject: [Disksim-users] Disksim migration In-Reply-To: <20061010233631.66429.qmail@web34012.mail.mud.yahoo.com> References: <20061010233631.66429.qmail@web34012.mail.mud.yahoo.com> Message-ID: <4fc1e0430610101642r7498c05dv4740a8a0e8faf475@mail.gmail.com> Try 'apt-get install flex-old' -- I believe there may be a problem with newer versions of flex. john On 10/10/06, shay raz wrote: > Disksim users, > I would like to migrate my disksim from old red-hat 7.3 to a newer platform > like ubuntu. > I installed on my ubuntu box gcc-2.95 and other tools like flex, bison etc. > Yet, I got the following error, trying to "make" disksim: > > lex.libparam.o lex.libparam.c > libparam.lex: In function `libparamlex': > libparam.lex:141: `yy_current_buffer' undeclared (first use in this > function) > libparam.lex:141: (Each undeclared identifier is reported only once > libparam.lex:141: for each function it appears in.) > libparam.lex: At top level: > lex.libparam.c:1315: warning: `yyunput' defined but not used > make[1]: *** [lex.libparam.o] Error 1 > > is that a version problem or some tool is missing? > Is error in libparam.lex related to flex package? > What should I do? > Very best, > shay > > > ________________________________ > Want to be your own boss? Learn how on Yahoo! Small Business. > > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > From mitzytest at gmail.com Thu Oct 12 15:49:08 2006 From: mitzytest at gmail.com (Nuengwong Tuaycharoen) Date: Thu, 12 Oct 2006 15:49:08 -0400 Subject: [Disksim-users] separate write cache Message-ID: Hi, I'm using Disksim 2.0. I am just not sure how to implement a separate write cache buffer from a read cache. I want my write cache buffer to (1) have multi-segments, (2) piggybacks the write requests when the read/write head passes the disk sectors, and (3) sends the write requests to the disk when the disk is idle. How can I set the parameter files to implement that? Right now I set these parameters: Number of buffer segments: 1024 Maximum number of write segments: 512 Use separate write segment: 1 Immediate buffer write: 1 The write cache papers claimed that the cache can improve the performance significantly. However, the output from my settings seems not to change much. Do I need to set anything else? Thank you. -- NT -------------- next part -------------- An HTML attachment was scrubbed... URL: From bucy at ece.cmu.edu Fri Oct 13 14:14:16 2006 From: bucy at ece.cmu.edu (John Bucy) Date: Fri, 13 Oct 2006 11:14:16 -0700 Subject: [Disksim-users] separate write cache In-Reply-To: References: Message-ID: <4fc1e0430610131114pdba0e34h934303c671dfe379@mail.gmail.com> Our recent experience has been that the disk cache code in disksim is pretty opaque and fragile; we've had instances where, e.g. changing the parameters to try to disable the writeback cache has uncovered bugs or otherwise didn't work. Sorry I can't be more helpful than that. john On 10/12/06, Nuengwong Tuaycharoen wrote: > Hi, > I'm using Disksim 2.0. I am just not sure how to implement a separate write > cache buffer from a read cache. I want my write cache buffer to (1) have > multi-segments, (2) piggybacks the write requests when the read/write head > passes the disk sectors, and (3) sends the write requests to the disk when > the disk is idle. How can I set the parameter files to implement that? Right > now I set these parameters: > > Number of buffer segments: 1024 > Maximum number of write segments: 512 > Use separate write segment: 1 > Immediate buffer write: 1 > > The write cache papers claimed that the cache can improve the performance > significantly. However, the output from my settings seems not to change > much. Do I need to set anything else? > > Thank you. > > -- NT > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > From raysmile at tom.com Mon Oct 30 20:50:50 2006 From: raysmile at tom.com (=?gb2312?B?zO/A2g==?=) Date: Tue, 31 Oct 2006 09:50:50 +0800 Subject: [Disksim-users] why does some disk scheduling algorithms not work in the DiskSim 3? Message-ID: <200610310950447030514@tom.com> Hi, I download and install the DiskSim 3 in my Redhat 9 box, and it can be compiled and run "valid/runvalid" well. But if I only change the IOdriver's Scheduling Policy from "1"(FCFS) to "8"(SPTF), it is aborted. why? Someone can provide some information? Lei 2006-10-31 -------------- next part -------------- An HTML attachment was scrubbed... URL: