herb@ajfcal.uucp (Herb Peyrel) (05/13/91)
Last week I upgraded my VERY old version of AIX 2.2.1 (I can't remember which release it was, but it was straight out of the binders) to version 1773 which I guess was IBM's latest offering. I'm having the following problems: (Which I did not have before) a) I have a Telebit TrailBlazer+ on a 4-port async card. For some reason, the card refuses to do any sort of hardware handshaking at all. It would in my previous version, however, after re-install (after making sure the port settings were identical), the hardware handshaking seems to have gone away. I now use Xon/Xoff which is resulting in grim transfer times. b) Same TrailBlazer, for some reason, "getty" never stays around very long. It always spawns login instantly and then a minute later, login dies leaving getty which then spawns another login... Ad-infinitum. My wtmp grows by about 100k/day.. The worst part is, it's very hard for my users to dial up to my site. c) I have a slip line to an Apollo DN3000 running at 19.2k. This worked perfectly before the upgrade, yet now when I telnet to the RT and try to use any application that requires curses, the telnet hangs. It seems to spit out some control characters, an 'H', more control characters, a '-', more control characters, and an 'E'... Waiting about 1 minute, it repeats this sequence, and after another minute comes up fine. This happens to "ate", "vi", "nn", and just about any other full screen application I can think of.. d) Async performance has just gone down the tubes too. I used to be able to have my Telebit bringing in news and use telnet on my slip line simultaneously. Now, if I watch the modem lights, the receive will be on solid and I'll be transmitting Xon/Xoff's regularly (see problem "a"), then as soon as I open a telnet from my Apollo, the Telebit stops sending any- thing. As soon as I exit the telnet session, the UUCP transfer continues. e) I have 11432 of PgSpace set up and my machine has 3 MB's of RAM. (I know, I know). However, I'm presently using ate to dial up my UUCP neighbor while compiling cnews in a telnet session on my Apollo... This is something I could reasonably do earlier, now the system reports low PgSpace and enters sync-mode almost constantly... It's quite annoying. It's getting so I can't do more than one thing at a time. If I wanted this, I'd use MS-Douche. (I'd love to increase my page space but that requires a re-install and requires me borrowing a tape-drive again, and that's quite difficult for me.). f) Not really a problem, but I wonder if anyone has drivers for any 3Com cards for AIX... I don't have any money to be buying an IBM Ethernet Adaptor but I have an extra 3c501, and 3c505 kicking around... I could easily get a 3c503 as well but would like to get a driver for it... Also, is there a SCSI driver for AIX that will work with a Western Digital WD7000 controller? My kingdom for AOS4.3... Anyways, please reply via E-Mail to herb@ajfcal.UUCP. My machine (herb@blender.UUCP) may be undergoing transformations in the near future so I am reading mail and some news here. Thanx in advance for any help you can provide. (Incidentally, I did the upgrade to fix the restore bug so I could install optional software.)
robin@pensoft.uucp (Robin Wilson) (05/17/91)
In article <1991May13.131804.1265@ajfcal.uucp> herb@ajfcal.uucp (Herb Peyrel) writes: >Last week I upgraded my VERY old version of AIX 2.2.1 (I can't >remember which release it was, but it was straight out of the >binders) to version 1773 which I guess was IBM's latest offering. > > a) I have a Telebit TrailBlazer+ on a 4-port async > card. For some reason, the card refuses to do any > sort of hardware handshaking at all. It would in my > previous version, however, after re-install (after making > sure the port settings were identical), the hardware > handshaking seems to have gone away. I now use Xon/Xoff > which is resulting in grim transfer times. AIX 2.2.1 has NEVER done RTS/CTS handshaking. So... How did you get it working before? Further, if you are doing file transfer, XON/XOFF is not supported. This is because most versions of file transfer use either CRC or checksums to verify data integrity. The CRC (or CheckSum) is a binary value that can happen to be the same as a valid flow control character (^S or ^Q). If you are doing XON/XOFF flow control, these CRC's will be stripped of the binary value of the XON or XOFF characters, and hence produce an invalid CRC. When the receiving side of the file transfer requests a resend, surprise the CRC is the same, and still contains the flow control character, and is still invalid. This will show up as a random failure in the transfer of certain files. (Although it show fail in the same place in the failing files, it will not neccessarily fail on all files.) > b) Same TrailBlazer, for some reason, "getty" never stays > around very long. It always spawns login instantly and > then a minute later, login dies leaving getty which then > spawns another login... Ad-infinitum. My wtmp grows by > about 100k/day.. The worst part is, it's very hard for > my users to dial up to my site. Try setting carrier detect to "follow true carrier". Mail me directly and I will provide you with a complete description of how each type of getty works and why carrier detect is SO significant. > c) I have a slip line to an Apollo DN3000 running at 19.2k. > This worked perfectly before the upgrade, yet now when I > telnet to the RT and try to use any application that requires > curses, the telnet hangs. It seems to spit out some > control characters, an 'H', more control characters, a '-', > more control characters, and an 'E'... Waiting about 1 minute, > it repeats this sequence, and after another minute comes up > fine. This happens to "ate", "vi", "nn", and just about any > other full screen application I can think of.. Call IBM Level 2 Defect Support (1800-237-5511). There is a fix for this problem. Make sure that your TERM is being negotiated to some type other than "hft", or "ibmxxxx" (where xxxx = 5081, 6153, 6154, etc.). > d) Async performance has just gone down the tubes too. I used > to be able to have my Telebit bringing in news and use telnet > on my slip line simultaneously. Now, if I watch the modem > lights, the receive will be on solid and I'll be transmitting > Xon/Xoff's regularly (see problem "a"), then as soon as I > open a telnet from my Apollo, the Telebit stops sending any- > thing. As soon as I exit the telnet session, the UUCP transfer > continues. What makes you say that you are transmitting "XOFF" or "XON"? Are you watching the line with a serial line analyzer? My guess is that this is a problem caused by a low number for "CHARLISTS" in your kernel. Edit the /etc/master file and set charlists to 1024 or some relatively large number. Again E-MAIL me directly for a more complete description. > e) I have 11432 of PgSpace set up and my machine has 3 MB's of > RAM. (I know, I know). However, I'm presently using ate to > dial up my UUCP neighbor while compiling cnews in a telnet session > on my Apollo... This is something I could reasonably do earlier, > now the system reports low PgSpace and enters sync-mode almost > constantly... It's quite annoying. It's getting so I can't do > more than one thing at a time. If I wanted this, I'd use MS-Douche. > (I'd love to increase my page space but that requires a re-install > and requires me borrowing a tape-drive again, and that's quite > difficult for me.). This could be caused by generating a high number of interupts on the serial port, but I am not sure. I suggest you take a closer look at what might be running on the system whne this happens. Does this problem clear up after you re-IPL the system? > f) Not really a problem, but I wonder if anyone has drivers for > any 3Com cards for AIX... I don't have any money to be buying > an IBM Ethernet Adaptor but I have an extra 3c501, and 3c505 > kicking around... I could easily get a 3c503 as well but would > like to get a driver for it... Also, is there a SCSI driver > for AIX that will work with a Western Digital WD7000 controller? Can't help you here. By the way, I only responded here to all of these questions to make sure an announce that mpp1779 updates are available from Defect Support. +-----------------------------------------------------------------------------+ |The views expressed herein, are the sole responsibility of the typist at hand| +-----------------------------------------------------------------------------+ |UUCP: pensoft!robin | |USNail: 701 Canyon Bend Dr. | | Pflugerville, TX 78660 | | Home: (512)251-6889 Work: (512)343-1111 | +-----------------------------------------------------------------------------+