The deadlock described below is caused by the dynamic allocation of memory in the TCP/IP stack. When you say DASD over Ethernet you actually are referring to DASD over TCP/IP on Ethernet. I assume that the DASD driver does not dynamically allocate memory in the read/write path. One might actually be able to dynamically allocate memory in the read path now that I think about it.
-----Original Message----- From: Linux on 390 Port [mailto:[EMAIL PROTECTED] Behalf Of shogunx Sent: Tuesday, November 15, 2005 9:36 AM To: LINUX-390@VM.MARIST.EDU Subject: Re: DASD over ethernet. On Tue, 15 Nov 2005, Carsten Otte wrote: > shogunx wrote: > > Thats the only standpoint I need... The 390 is a R&D environment, not > > used in production, and we own it outright. SO, theoretically, I could > > slap a larger hd in the console of the 390, run linux on it, telnet into > > the 390, and mount a partition of the drive in the console as the > > DASD via network block device, thereby eliminating the need for an > > external SAN to provide DASD? I know, I lose redundancy, etc, but I can't > > seem to acquire code to operate the SSA SerialRAID controllers in my > > Shark, which I ported linux to. > Interresting project to port Linux to your shark ;-). I had to. I couldn't get higher than "service" priveleges on the shark hosts to be able to do anything with it when I bought it used, so my workaround was yank the 9gb system disk, put a filesystem and kernel on in(gentoo with a hand tweaked kernel compiled on a 7025 running SuSE), put it back in the shark, and boot. loop that a few times for trial and error on the kernel/fs and it worked. they are nice hosts, as I'm sure you know. lack of code for the SSA subsystem and PCI expansion enclosures is a bit of a pain, but better than no access to the host. > It *should* work as > far as I can tell, but be aware that you can run into nasty deadlocks once > you don't have a local swap disk: you'll need networking for paging/disk IO, > and once linux is hard out of memory networking is suspended until there is > memory available => deadlock. How would that be different than having swap on standard SAN housed DASD? Does ficon/escon not suspend under those conditions? > Workaround could be to have a ramdisk or z/VM DCSS as swap target, or to go > to church once a week ;-). Ramdisk would probably do the job. I'll start by installing linux on the bare metal of the 9672. BTW, there are only 5 processors live on this particular 9672, I suppose I would have to pay to get the other 7 functional? Thanks, Scott > > cheers, > Carsten > -- > > Carsten Otte > IBM Linux technology center > ARCH=s390 > sleekfreak pirate broadcast http://sleekfreak.ath.cx:81/ ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390