On 09/17/2011 12:08 AM, Mike Christie wrote: > On 09/16/2011 09:01 AM, Heinrich Langos wrote: >> What it boils down to is no significant effect of iptables ! >> > > Ok. Different issues maybe then. Let me do some more digging. I have to > kill some other regression that I have been working on for work (also > seen in that "Re: open-iscsi issue" thread) then I can concentrate more > on this. > > The strange thing is that I added the iscsi code from upstream to rhel > 6/5 and it is working fine with or without iptables for me. I get 113 > MB/s read or write on 1 gig. > > Is your box 32 bit or 64 and if 32 bit are you using highmem? > > If it is 64 bit and you have time and like to build kernels try and do a > "git bisect" to narrow down what kernel commit is causing the problem. >
I would like to report that with 3.0 Kernel I can do 103MB on a single osd iscsi-device. osd is very different but it uses the exact same iscsi-LLD, only the upper layers above scsi are completely different. Could you also test with direct IO? If you have time, download the "sg utils" (search on google) and use the sg_dd command with direct IO use big block sizes of 8MB or so. With this we can see if it comes from Low-Level-Driver or its an upper layer issue. Because even if accessing a block device directly dd is still using page_cache. Also what about an ext4 FS ontop of the iscsi device, and dd into a file. Is that the same results? Thanks Boaz -- You received this message because you are subscribed to the Google Groups "open-iscsi" group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
