4096 is the PG_SIZE kernel param. getconf PAGE_SIZE As you correctly pointed out, it's not related to s3ql. Small question: tunning the kernel to use bigger PG_SIZE value (e.g. 65536) and formatting the partition where s3ql cache resides with 65536 block size will bring improve s3ql performance?
On Thursday, August 21, 2014 1:21:02 PM UTC-6, Dinu Cosmin wrote: > > Hi, > > My setup consist in a s3ql mount exported by nfs. The max-object-size > value used in mkfs.sql3 was 128 MB > > On the nfs client, the /proc/mount shows rsize=524288,wsize=524288 > meaning the nfs read and write block sizes are 512K. > > > Now when I copy a file over the network to the nfs mount, the > .s3ql/mount.log on the nfs server shows block size of 4K > > 2014-08-21 12:52:15.519 [pid=19104, thread='Dummy-23', module='s3ql.fs', > fn='write', line=1037]: write(4, 10452992, datalen=*4096*): start > > 2014-08-21 12:52:15.519 [pid=19104, thread='Dummy-23', module='s3ql.fs', > fn='write', line=1037]: write(4, 10457088, datalen=*4096*): start > > 2014-08-21 12:52:15.520 [pid=19104, thread='Dummy-23', module='s3ql.fs', > fn='write', line=1037]: write(4, 10461184, datalen=*4096*): start > > > My question is why the write block size is so small e.g. 4K? How can I > increase the existing 4096 block size used in write? > > > Thanks for the answer! > -- You received this message because you are subscribed to the Google Groups "s3ql" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
