I'm pleased to see that the new release has full support for asynchronous I/O. I was surprised, though, since I thought that the aio bug in the 2.4 kernels couldn't be fixed without breaking binary compatibility. Was there a special fix for this in the RHEL3 update 4 kernel? Anyway, thanks for all your efforts. We're looking forward to testing aio with the new version next week. Derek -------------- next part -------------- An HTML attachment was scrubbed... URL: http://oss.oracle.com/pipermail/ocfs-users/attachments/20050131/02e321ad/attachment.html
the symbols we needed were not exported in the kernels, so we got red hat to export the symbols on rhas21 e57 and rhel3 update4. since it's just exportingsymbols, no breakage in abi. and we ended up doing the work inside ocfs rather than relyign on the aio layer to handle non contig io On Mon, Jan 31, 2005 at 11:13:15AM -0800, Derek Suzuki wrote:> I'm pleased to see that the new release has full support for asynchronous I/O. I was surprised, though, since I thought that the aio bug in the 2.4 kernels couldn't be fixed without breaking binary compatibility. Was there a special fix for this in the RHEL3 update 4 kernel? > Anyway, thanks for all your efforts. We're looking forward to testing aio with the new version next week. > > Derek> _______________________________________________ > Ocfs-users mailing list > Ocfs-users@oss.oracle.com > http://oss.oracle.com/mailman/listinfo/ocfs-users
In 1.0.13, we were using the generic aio r/w calls (generic_file_aio_read, etc.) provided by the kernel. Instead of fixing those and deal with the ABI compatibility issues, we implemented them afresh as ocfs_aio_read/write. To do that, we needed few unexported symbols exported. The EL3 U4 exposes those symbols to the modules. Let us know how your testing goes. Sunil Derek Suzuki wrote:> I'm pleased to see that the new release has full support for > asynchronous I/O. I was surprised, though, since I thought that the > aio bug in the 2.4 kernels couldn't be fixed without breaking binary > compatibility. Was there a special fix for this in the RHEL3 update 4 > kernel? > > Anyway, thanks for all your efforts. We're looking forward to > testing aio with the new version next week. > > Derek > >------------------------------------------------------------------------ > >_______________________________________________ >Ocfs-users mailing list >Ocfs-users@oss.oracle.com >http://oss.oracle.com/mailman/listinfo/ocfs-users > >
Hello, I read these posts with interest, but I can't locate the specifics on the AIO kernel bugs your talking about. Do you have any links/specifics on those issues you brought up? I am wondering if we are seeing behavior on one of our databases that could be related. Our Database RAC node will cause the database to "freeze" momentarily when that node is archiving. The DB archives to an Ext2 partition and the DB is OCFS. Thanks Bruce H.