jb@aablue.UUCP (John B Scalia) (11/08/89)
My ESIX saga continues...
I decided to rebuild my entire system, because, *sigh* the boss wanted a
Dos partition on the drive as well :-(. Anyway, this time I decided to
go a little further on the installation sequence and add TCP/IP, RFS,
& X-windows, just to see how slow it really was. [see previous threads
about ESIX & X-windows]
Installation was simple and uneventful so I knew I was in trouble. The
problem developed immediately after I finished the installation and
tried to uucp some routines from the office mini. I had not taken the time
to connect the system to the office ethernet or even to install the
E-net card in the PC. The unit was direct connected by a serial line.
At this point, my HDB uucp stuff, which had been operating perfectly,
became dazed & confused. While I managed to get the routines I needed,
my throughput dropped down to approx. 230cps. This was also the case
when I used the TB+ modem with an outside site. Thinking, perhaps, I
had futzed something during the install, I used "rmpackage" and brought
the unit back to its old configuration. The HDB was back up to work
and showed throughput in the range of 1700cps after the net stuff
was uninstalled.
The ESIX people have been of no help; surprise. Is there something in
the TCP stuff I need to look for? I know squat about networking
software, but I learn fast if I know where to look. Could the problem
go away when I connect it to the office net? It takes a little while
to get cables put in here, but that might happen next week, if I
scream loud enough.
Humbly awaiting suggestions (& followups to comp.unix.i386 please)
jb@aablue
--
A A Blueprint Co., Inc. - Akron, Ohio +1 216 794-8803 voice
UUCP: {uunet!}aablue!jb (John B. Scalia)
Just a little more nonsense to clutter up the net.palowoda@fiver.UUCP (Bob Palowoda) (11/12/89)
From article <612@aablue.UUCP>, by jb@aablue.UUCP (John B Scalia): > My ESIX saga continues... > > I decided to rebuild my entire system, because, *sigh* the boss wanted a > Dos partition on the drive as well :-(. I run dos programs with ESIX on Simul-task on the unix partition. Why make a dos partition? [a little stuff deleted] > At this point, my HDB uucp stuff, which had been operating perfectly, > became dazed & confused. While I managed to get the routines I needed, > my throughput dropped down to approx. 230cps. This was also the case > when I used the TB+ modem with an outside site. Thinking, perhaps, I > had futzed something during the install, I used "rmpackage" and brought > the unit back to its old configuration. The HDB was back up to work > and showed throughput in the range of 1700cps after the net stuff > was uninstalled. Interesting, I installed the tcp-ip and it didn't affect my uucp throughput (TB+ modem) at all. What baudrate was it set at after the the tcp install? > > The ESIX people have been of no help; surprise. Is there something in > the TCP stuff I need to look for? I know squat about networking > software, but I learn fast if I know where to look. Could the problem > go away when I connect it to the office net? It takes a little while > to get cables put in here, but that might happen next week, if I > scream loud enough. Check to see if you have the proper /dev/ttyMxx lines after you install the tcp-ip package. If not add them to your /etc/conf/node.d/asy file and make the proper changes in /etc/conf/cf.d/init.base file. It had to tell what's going on from the information you gave. I'm suprised that you get 1700cps from your TB+. Sounds like your lines are too perfect. Are these numbers from transfering files greater than lets say 50K bytes? ---Bob -- Bob Palowoda pacbell!indetech!palowoda *Home of Fiver BBS* login: bbs Home {sun|daisy}!ys2!fiver!palowoda (415)-623-8809 1200/2400 Work {sun|pyramid|decwrl}!megatest!palowoda (415)-623-8806 2400/9600/19200 TB Voice: (415)-623-7495 Public access UNIX XBBS
peter@ficc.uu.net (Peter da Silva) (11/15/89)
What TCP/IP does ESIX come with? In article <612@aablue.UUCP> jb@aablue.UUCP (John B Scalia) writes: [serial port performance really amazingly bad with TCP up] > The ESIX people have been of no help; surprise. Is there something in > the TCP stuff I need to look for? What run level did you install the TCP at? I always follow convention and put networking at run level 3. Then I edit /etc/conf/init.d/whatever to set the default run level to 2. That way you can see the effect of active TCP instead of passive stuff. I have a 'neton' login, no password, that does a telinit 3. With Belltech/Intel V.3.2u, with Lachman TCP/IP: when I have TCP running the performance of the serial port drops dramatically, and I get about 1 in 50 characters lost even at typing speeds. > I know squat about networking > software, but I learn fast if I know where to look. Could the problem > go away when I connect it to the office net? Hasn't for me. You might try playing with the run level as I suggested. -- `-_-' Peter da Silva <peter@ficc.uu.net> <peter@sugar.hackercorp.com>. 'U` -------------- +1 713 274 5180. "*Real* wizards don't whine about how they paid their dues" -- Quentin Johnson quent@atanasoff.cs.iastate.edu
cpcahil@virtech.uucp (Conor P. Cahill) (11/15/89)
In article <6965@ficc.uu.net>, peter@ficc.uu.net (Peter da Silva) writes: > [serial port performance really amazingly bad with TCP up] > > With Belltech/Intel V.3.2u, with Lachman TCP/IP: when I have TCP running > the performance of the serial port drops dramatically, and I get about > 1 in 50 characters lost even at typing speeds. I am running 386/ix with TCP, NFS, X, etc. and have no problems with serial port performance with the networking up or not. I do have an intelligent serial port card which may be the reason. -- +-----------------------------------------------------------------------+ | Conor P. Cahill uunet!virtech!cpcahil 703-430-9247 ! | Virtual Technologies Inc., P. O. Box 876, Sterling, VA 22170 | +-----------------------------------------------------------------------+
dave@pmafire.UUCP (Dave Remien) (11/16/89)
In article <1989Nov15.133059.3330@virtech.uucp> cpcahil@virtech.uucp (Conor P. Cahill) writes: >In article <6965@ficc.uu.net>, peter@ficc.uu.net (Peter da Silva) writes: >> [serial port performance really amazingly bad with TCP up] >> >> With Belltech/Intel V.3.2u, with Lachman TCP/IP: when I have TCP running >> the performance of the serial port drops dramatically, and I get about >> 1 in 50 characters lost even at typing speeds. > [Conor say 386/IX is similar] I'm running Bell Tech 3.2 (NOT 3.2u, I'm waiting for that) with Lachman TCP/IP, NFS, X and both a regular serial port and a Bell Tech ACE card; I have yet to see any slowdown yet that can't be attributed to a 16MHz '386 and only 4 megs of RAM. I've got an ENET586 card, for what it's worth. -- Dave Remien ++*+*+*+*+*+*+*+*+*+*+*+*+*+* WINCO Computer Engineering Group {uunet|bigtex}!pmafire!dave or rzd@inel.gov "Dave Barry for President"
seg@stubby.UUCP (Scott E. Garfinkle) (11/16/89)
From article <6965@ficc.uu.net>, by peter@ficc.uu.net (Peter da Silva): > In article <612@aablue.UUCP> jb@aablue.UUCP (John B Scalia) writes: > [serial port performance really amazingly bad with TCP up] > >> The ESIX people have been of no help; surprise. Is there something in >> the TCP stuff I need to look for? The ESIX people were pretty helpful with me. I called up (I have never yet had to wait on hold), asked my question, then called back the next day. They told me that there seems to be some kind of bug in sendmail that kills the system. I haven't had a chance to test this yet (I won't be back home for a couple weeks), but it sounds like a reasonable explanation. I was planning to use smail, anyway. My only complaint so far is that, contrary to the implication of their ads, NFS is *not* included with ESIX. It turns out to be a $550 optional extra, and doesn't even come with YP support! -scott garfinkle
johnl@esegue.segue.boston.ma.us (John R. Levine) (11/17/89)
In article <862@pmafire.UUCP> dave@pmafire.UUCP (Dave Remien) writes: >[many people report lousy serial performance on various 386 unices] >I'm running Bell Tech 3.2 (NOT 3.2u, I'm waiting for that) with Lachman >TCP/IP, NFS, X and both a regular serial port and a Bell Tech ACE card; >I have yet to see any slowdown yet that can't be attributed to a 16MHz >'386 and only 4 megs of RAM. I've got an ENET586 card, for what it's >worth. It depends mostly on the controller chip in your serial card. The problem is that the 386 doesn't really support priority interrupts. (There is a global suppress interrupt bit, the priority business on the 8259 only applies when several interrupt requests arrive at once.) Too many device drivers disable interrupts while doing non-trivial processing, so that if you're trying to run at 19.2K and interrupts are off for 500us or more, you lose. If you can replace your serial chip with a 16550, there are several serial drivers that turn on the silo in the 16550, greatly relaxing the interrupt response required. The 386/ix X5 update driver does this, as does the free JJM driver that's been passed around. The right solution would be to be more disciplined about driver interrupt handling, and to put in a kernel mechanism to let drivers schedule work "later", meaning not at interrupt level but before returning to the mainline user or kernel process. I expect that's what Venix does, but don't hold your breath waiting for the regular Unix vendors to do it -- it'd be a lot of work. -- John R. Levine, Segue Software, POB 349, Cambridge MA 02238, +1 617 864 9650 johnl@esegue.segue.boston.ma.us, {ima|lotus|spdcc}!esegue!johnl "Now, we are all jelly doughnuts."
peter@ficc.uu.net (Peter da Silva) (11/17/89)
[I complained of slow serial performance with Lachman TCP running] > I am running 386/ix with TCP, NFS, X, etc. and have no problems with > serial port performance with the networking up or not. I do have an > intelligent serial port card which may be the reason. I dare say it is... I'm using COM1. -- `-_-' Peter da Silva <peter@ficc.uu.net> <peter@sugar.hackercorp.com>. 'U` -------------- +1 713 274 5180. "vi is bad because it didn't work after I put jelly in my keyboard." -- Jeffrey W Percival (jwp@larry.sal.wisc.edu)
cpcahil@virtech.uucp (Conor P. Cahill) (11/17/89)
In article <862@pmafire.UUCP>, dave@pmafire.UUCP (Dave Remien) writes: > In article <1989Nov15.133059.3330@virtech.uucp> cpcahil@virtech.uucp (Conor P. Cahill) writes: > >In article <6965@ficc.uu.net>, peter@ficc.uu.net (Peter da Silva) writes: > >> [serial port performance really amazingly bad with TCP up] > >> > >> With Belltech/Intel V.3.2u, with Lachman TCP/IP: when I have TCP running > >> the performance of the serial port drops dramatically, and I get about > >> 1 in 50 characters lost even at typing speeds. > > > [Conor say 386/IX is similar] I did not say 386/ix was similar. I said: I am running 386/ix with TCP, NFS, X, etc. and have no problems with serial port performance with the networking up or not. I do have an intelligent serial port card which may be the reason. Which in other words says that I do not have the same problem. > I'm running Bell Tech 3.2 (NOT 3.2u, I'm waiting for that) with Lachman > TCP/IP, NFS, X and both a regular serial port and a Bell Tech ACE card; > I have yet to see any slowdown yet that can't be attributed to a 16MHz > '386 and only 4 megs of RAM. I've got an ENET586 card, for what it's > worth. However both my system and yours use an intelligent serial card for serial communications and this may be the key. The original posters did not indicate the kind of serial port they were using, but I would bet that it is the standard dump port. -- +-----------------------------------------------------------------------+ | Conor P. Cahill uunet!virtech!cpcahil 703-430-9247 ! | Virtual Technologies Inc., P. O. Box 876, Sterling, VA 22170 | +-----------------------------------------------------------------------+