Hmm, I wasn't previously aware of /etc/sysconfig/harddisks... I had just been calling hdparm from rc.local.... good to know.
Unfortunately, that doesn't really solve my problem... I'm curious, what features did you have enabled which resulted in data corruption? the man page warns of corruption with unmaskirq, so I haven't tried it, but I've seen it enabled on other systems by default with no harm done, and I suspect it would provide a performance boost for my application. -----Original Message----- From: Stephen Kuhn [mailto:[EMAIL PROTECTED] Sent: Wednesday, March 19, 2003 11:50 AM To: [EMAIL PROTECTED] Subject: Re: default hdparm settings On Thu, 2003-03-20 at 06:37, Kevin Bowen wrote: > Hi, > I was wondering if anyone knows what the story is with how redhat decides what > default hard disk settings (i.e. hdparm settings for DMA, 32 bit IO, unmaskirq etc) > to use on a new install? I've done several installs, and on some of them I've seen > it default to all features turned on, other times, all features off. Does redhat > installer have a blacklist, or a whitelist of drive/controller combo's on which it > will enable certain features? Or are these variations bugs? On the systems where > everything is turned off by default, I've manually enabled DMA and 32bit, but the > man page entry for unmaskirq scares me a bit. Does anyone know of any resource for > determining the safety of various settings for various drives/controllers? > > ************************************************************************ > Kevin Bowen > UCSD ACS/Network Operations, Customer Support Help Desk > [EMAIL PROTECTED] / (858) 534-1857 / http://www-no.ucsd.edu > Phone Support: 7 AM to 7 PM / Walk-in Support: 10 AM to 3:30 PM > ************************************************************************ By default, I always check the /etc/sysconfig/harddisks file to see what's been either turned on, or left off; this is where you can specify your "tuning" params in an easy way. As of yet, I've never been impressed with the EIDE performance "out of the box", so I tend to pass params to the kernel as well as tweak the settings of the /etc/sysconfig/harddisks file. On RH 8.0, I have experienced some problems with the combination of the two - ending with corruption and hosed performance. Otherwise, it's been a good combo for me over the years. HTH -- Thu Mar 20 06:45:00 EST 2003 06:45:00 up 1 day, 8:30, 4 users, load average: 0.07, 0.11, 0.09 ------------------------------------------------------------------ | __ __ | kuhn media australia | | / ,, /| |'-. | http://kma.0catch.com | | .\__/ || | | |=================================| | _ / `._ \|_|_.-' | stephen kuhn | | | / \__.`=._) (_ | email: [EMAIL PROTECTED] | | |/ ._/ |"""""""""| | email: [EMAIL PROTECTED] | | |'. `\ | | | icq: 5483808 | | ;"""/ / | | | | | smk ) /_/| |.-------.| | mobile: 0410-728-389 | | ' `-`' " " | Berkeley, New South Wales, AU | ------------------------------------------------------------------ linux user:267497 * MDK 9.1 * PC/Mac/Linux/Networking/Consulting machine no:194239 * RH 7.3 * Sales - Service - Support - Tutor ------------------------------------------------------------------ ** This messages was composed on a 100% Microsoft free computer ** A mathematician, a doctor, and an engineer are walking on the beach and observe a team of lifeguards pumping the stomach of a drowned woman. As they watch, water, sand, snails and such come out of the pump. The doctor watches for a while and says: "Keep pumping, men, you may yet save her!!" The mathematician does some calculations and says: "According to my understanding of the size of that pump, you have already pumped more water from her body than could be contained in a cylinder 4 feet in diameter and 6 feet high." The engineer says: "I think she's sitting in a puddle." -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/redhat-list -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/redhat-list
