. |
RE: XPostFacto 3 boot problems |
September, 08, 2003 12:31 AM |
nick.ashton |
. |
powderhaus - before you rush out to buy a Sonnet FW/USB card bear in mind that I'm already using one and seeing the same problems though maybe not to the extent that you are. |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 11:58 PM |
mjoecups358 |
. |
Your complaining about 20M per sec? Get real! That really good for a single drive. Sheesh, give an inch and the expect a mile :~> Marty |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 8:59 PM |
powderhaus |
. |
I just checked Bugzilla about the issue, you answer it already, thanks. Looks like i am off to buy a sonnet card, hopefully the transfur rates are better too, this card tops out at 20-23MB/s. thanks again, and i also agree that bugzilla will be very cool. |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 8:18 PM |
powderhaus |
. |
The drive it self has a Oxford 911 chipset, the Belkin card has always seemed to be a problem but i never had the "smoking gun" because it started acting up when i got my firewire HD, but that was my second FW device. I have no idea what chipset teh Belkin card is.... Ryan, is there anyway you can update the compatability page to include what FW and FW/USB cards work the best? I need a FW/USB card because i can only have 2 PCI devices and my video card is on of them. Glad to see you back! I had to reinstall OSX again because there was a problem with my FW drive (a currupt block or something, because XPF could not write to it and before that it would crash at the grey screen and never boot up. Umax J700 G4 400 664MB RAM ATI Radeon Belkin FW/USB card Generic FW case (Oxford 911) Yamaha CDRW Logitech USB keyboard (sucks that the USB stuff is not supported at boot) SmartCard USB reader. |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 2:11 PM |
OSXGuru |
. |
Yes, this is exactly what I get when trying to boot from problematic Firewire PCI cards. I've opened a bug on it--here's the URL: http://www.opendarwin.org/bugzilla/show_bug.cgi?id=879 |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 12:33 PM |
chibi_delenn |
. |
Just a thought - it might or might not be relevant to the situation, but you might want to post what FW bridge chipset (Oxford 911, 921, etc) you are using. That may make a big difference in terms of stability. Just a longshot in the dark, but you never know... Chibi Delennâ„¢ |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 11:50 AM |
powderhaus |
. |
My system seem to boot fine with the exception of the 50 IOfirewire errors, but i get the "grey" screen of death right after it fiishes booting. It should proced to log me in but it seems to crash here without reporting any errors befcause verbose has just exited. I used to get some login error throttle - 24 Thanks for any help, Jim |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 10:22 AM |
nick.ashton |
. |
I have done some more experiments which seem to confirm that dodgy I/O to the Firewire drive is the reason for the boot problems. I had one particular service, VNC, which was failing to start properly every time during the boot sequence. So I moved its startup scripts to a SCSI disk and used a Unix link to reference this from the FW startup disk. Sure enough VNC started fine once these changes had been made. I have found that using the XPF Throttle feature seems to reduce, though not eliminate the I/O errors. Currently I'm using a throttle value of 8. Strangely, once booting is completed, the system seems to be very stable and I haven't had any unexpected errors which might be due to I/O problems. |
. |
RE: XPostFacto 3 boot problems |
September, 07, 2003 3:52 AM |
nick.ashton |
. |
Ryan has mentioned in the release notes for XPF 3 that there seems to be a problem with the OS X Firewire drivers running on old world machines which result in bad data being read from FW disks particularly during the boot sequence. He also stated that dedicated FW controller cards rather than combo cards seemed to work more reliably. Maybe it's a timing issue. Anyway, for what it's worth I have 10.2.6 running from a Firewire drive on my 8600 via a Sonnet Tango FW/USB combo card. BUT I am getting the following errors reported during the boot sequence IOFireWireSBP2Login::fetchAgentWriteComplete fetch agent write failed! retrying Usually I only see two or three of these, sometimes none. Also various processes that are scheduled to start during the boot sequence report errors such as process_name: (ipc/send) invalid destination port in the system log file (/var/log/system.log). Sometimes startup processes crash altogether, usually with a BAD_ACCESS or BAD_INSTRUCTION exception. Look in / Library/Logs/CrashReporter for evidence of this. I've found that it is essential to boot in verbose mode to ensure that the boot procedure completes but the random failure of startup processes means that often the boot ends at a blank grey screen or a blank blue screen. I have also experienced several cases where the login screen appears but the mouse cursor is frozen in the top left corner of the screen. If any of these happens to you then try pressing the power button on your keyboard twice followed by the return key (it must be an ADB keyboard). This should perform a clean shutdown of the system and avoid having to repair the mounted disks - much better than a forced reboot. I think all these bugs are down to the unreliable I/O to the Firewire boot disk. Hopefully we'll be able to narrow down the problem and get it fixed but since it may be in the OSX Firewire drivers themselves that could be tricky. It would help if people specified what machine, CPU, Firewire controller and Firewire disks they were using so that we can see if there is a pattern. My own details are 8600, G3/400/1MB, 352MB RAM, Sonnet Tango FW/USB card, LaCie 20GB FW drive |