Date: Mon, 13 Jul 1998 00:19:50 -0400 (EDT) From: Mikhail Teterin <mi@aldan.algebra.com> To: current@FreeBSD.ORG Subject: soft updates, ccd, old drives Message-ID: <199807130419.AAA08502@rtfm.ziplink.net>
next in thread | raw e-mail | index | archive | help
I'm not sure which part to blame, so just read it and make your own conclusion. I built the soft-updates kernel, and installed world last night. Then, I've decided to try it on my little ccd array (I did remove the async mount option first). The array consists of four disks: sd2: <CONNER CP30540 545MB3.5 B0BD> type 0 fixed SCSI 2 sd3: <CONNER CP30540 545MB3.5 B0BC> type 0 fixed SCSI 2 sd4: <CONNER CP30540 545MB3.5 AFBA> type 0 fixed SCSI 2 sd6: <CONNER CP30540 545MB3.5 AFBA> type 0 fixed SCSI 2 they are on a dual bus Adaptec : ahc0: <Adaptec aic7770 SCSI host adapter> at 0x9c00-0x9cff irq 10 ahc0: on eisa0 slot 9 ahc0: Using Level Sensitive Interrupts ahc0: aic7770 <= Rev C, Twin Channel, A SCSI Id=7, B SCSI Id=7, 4 SCBs ahc0: Resetting Channel B ahc0: Resetting Channel A ahc0: Downloading Sequencer Program...ahc0: 376 instructions downloaded two on one bus and two on the other. First weirdness is that kernel said "ffs_mountfs: superblock updated" every time I mounted my /ccd . It does not say that if I disable soft updates, so I report it. Now, `cd /ccd/ports; make clean' was working just fine (that's my place for the ports-tree). But extracting anything was hanging one disk or the other: with kernel messages about sd[46] "timeout in dataout" or similar. This would not respond to ctrl-alt-del, and, though the shells were working in other virtual consoles, they could not execute anything, including "reboot". So -- cold reset to the horrible fsck screenfulls. Hardware, I thought -- the poor old disks were never pushed that hard with the simple async mounts. So I grabbed the EISA floppy and set the maximum transfer rate from 8Mbps to the minimum 3.7Mbps. Same hangs, although this time there were no "timeout" messages on the console. I gave up, and disabled soft updates on /ccd (why is the tunefs option `-n'? did anyone notice, that the character 'n' is not even used in "soft updates"!). Since then, I successfully extracted XFree86 and gcc28 ports, and am currently building the gcc28 without problems. My conclusion is, that something in Soft Updates and CCD combination is wrong, which was hanging AND causing the disks to be pushed hard. After I took the maximum transfer rate down, the disks were not pushed as hard, so -- no timeouts, but the hangs were still here. This is a debug kernel and I have a corresponding kernel.debug. If someone wants me, I can try different things here to try to resolve this. Yours, -mi To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199807130419.AAA08502>