ncbergst@ndsuvax.UUCP (Troy Bergstrand) (07/14/88)
We just installed Emulex's UD33 disk controller on our VAX 780. But unfortunately the installation did not complete without incident. Our configuration: Vax 11-780 12 Megabytes of memory 2 Unibus controllers Unibus 1: UDA33 - 2 RA81, Interlan Ethernet Controller Unibus 2: 11 DMF32's, 1 DH We are running 4.3BSD Unix with no modifications to the uda or unibus code. We desired to put both the UD33 and the UDA50 on the same unibus, Emulex tells us this should work, but it didn't. After setting the UDA50 jumpers to the longest possible (10ms) DMA delay and setting the UD33 in the same way and also reconfiguring the kernel to place the UD33 at different unibus addresses we also got the same result. That result was a hanging of the system on disk i/o. I could run systat -vm and watch the disk requests and after a period of time there would arise some situation where at least one of the disk requests would not or could not be serviced. After which near every other process on the system would get stuck in the request queue behind the hung process. So to get the new drive working we put the UD33 on the unibus 2 with the communcation controllers and everything is ok there. Has anyone succesfully installed a UD33 on a 4.3BSD unix system on the same unibus as an UDA50? -- J. Troy Bergstrand North Dakota State University, Fargo, ND 58105 UUCP: ...!uunet!ndsuvax!ncbergst BITNET: ncbergst@ndsuvax.bitnet ARPA,CSNET: ncbergst%ndsuvax.bitnet@cunyvm.cuny.edu
chris@mimsy.UUCP (Chris Torek) (07/23/88)
In article <1030@ndsuvax.UUCP> ncbergst@ndsuvax.UUCP (Troy Bergstrand) writes: >We are running 4.3BSD Unix with no modifications to the uda or unibus code. >We desired to put both the UD33 [a UDA50 emulator] and the UDA50 on the >same unibus, Emulex tells us this should work, but it didn't. ... >result was a hanging of the system on disk i/o. ... I used to have a similar problem on a 780 with two UDA50s, and someone at Bell Labs had one with an 8600 with three of them. After installing my driver (which is included with 4.3BSD-tahoe) the problem disappeared. -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163) Domain: chris@mimsy.umd.edu Path: uunet!mimsy!chris
chris@mimsy.UUCP (Chris Torek) (07/24/88)
In article <12655@mimsy.UUCP> I wrote: >I used to have a similar problem [machine hangs waiting for I/O] on a 780 >with two UDA50s .... I should have mentioned that the problem sometimes appeared to be that the RA81s on one of the UDA50s would go off-line spontaneously. This can be seen as the `A' or `B' light on the drive going out. Other times, the UDA50 itself would hang. This shows as an interruption in the regular light pattern on one of the two Unibus modules (the usual pattern is one light on, one blinking; the blinking would stop, as I recall). Since then the microcode on the UDA50 has been revised. (Current microcode appears as `version 5 model 6' in printouts from my driver. I believe the off-line bug was fixed in version 4, and the hang bug in version 5, or something like that.) -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163) Domain: chris@mimsy.umd.edu Path: uunet!mimsy!chris