elee4fg@jetson.uh.edu (05/12/91)
From: seng Gan
Star Enterprise
Here are the problems that I known about:
I have AIX 3.1 running on 540 and 530.
540 and 530 each has 3 IBM 857MB internal harddisk and 2 external 1 GB
harddisk.
1) csh does not support suid. If your csh suid scripts file has this
#!/bin/csh
It won't work.
Your need to do this in ksh
/bin/csh <your suid script file> ( without #!/bin/csh )
2) tar to 2.3 GB 8mm drive
It is unreliable
I have tar 2 GB of files to the tape and backup
1,400 of 1,600 files. It skip files (200).
I have playing with this for 5 months. It does not works relialy.
You better check your tape carefully.
3) cpio
It skip files also.
4) printer attached to RS 422 port
forget it. It dies every 4 hours. The queue daemon went out to lunch
regularly.
We used LPD print server software from FTP software running on PC
as print server and use remote print queue.
5) Rs/6000 cannot telnet to SCO Xenix
But can rlogin to SCO Xenix
6) Increase your process limit
your need to edit /usr/include/sys/limits.h
CHILD_MAX to increase your number of process allowed per user id
in ksh 40 is too small for root
chdev -l sys0 -A maxuproc has no effect in ksh
7) There is some strange limit to number of device attached to SCSI card
Basically, if you have internal device attached to SCSI adapter
you can only have one or two extenal device per adapter. It applies in
540 and 530 also in addition to 320.
8) inetd
I let my rs/6000 PING EACH OTHER every hour. That seems to be the only
way to keep it running constantly.
9) PC/TCP telnet to RS/6000
If the session did not terminate gracefully, it drive RS/6000 nuts.
You will see sar command show system % time > user % time.
ps -ef will show pts/? session that did not show up in w command
ps au will show the session that eat away your CPU time.
Call the user, he/she will say I am not login.
kill the session, things will go back to normal.
10) RS/6000 (token ring TCP/IP) connect to Cisco
Need to define static ARP route in Cisco to RS/6000 or else the link
is not stable.
I will post more problem as I discover them.
Seng C, Gan
Unix SysAdm
Star Enterprise
Tel:(713)874-7816
Call me if you know any more problem about TCP/IP that I am not aware of.jona@iscp.Bellcore.COM (Jon Alperin) (05/13/91)
In article <9357.282caa94@jetson.uh.edu>, elee4fg@jetson.uh.edu writes: |> From: seng Gan |> Star Enterprise |> |> Here are the problems that I known about: < list trimmed > |> 10) RS/6000 (token ring TCP/IP) connect to Cisco |> Need to define static ARP route in Cisco to RS/6000 or else the link |> is not stable. We have found that any workstations running YP/NIS need to have routes defined statically, since AIX will set up these routes during a boot before getting the NIS information. Thus, if you use mnemonic names instead of addresses AIX cannot do an IP lookup against the host table (which it has not yet gotten from NIS), and therefore the route is lost. |> Seng C, Gan |> Unix SysAdm |> Star Enterprise |> Tel:(713)874-7816 -- Jon Alperin Bell Communications Research ---> Internet: jona@iscp.bellcore.com ---> Voicenet: (908) 699-8674 ---> UUNET: uunet!bcr!jona * All opinions and stupid questions are my own *
marc@stingray.austin.ibm.com (Marc J. Stephenson/140000;1C-22) (05/13/91)
(In reaction to an article describing several problems encountered...) When discussing problems with various levels of AIX, please be sure to identify the level of software which your machines have. There are various ways of doing that - one is by mentioning the service level (3003, 3005, etc) and another is by including in your submission the level of bos.obj (or boskj.obj for you Kanji folks), which can be found via the cmd "lslpp -h bos.obj". Though not a problem in the referenced article, if it is not clear from the content, please include the hardware platform you are running on. AIX for PS/2 is not the same as AIX for the '6000. Now, back to your regularly scheduled newsgroup.... -- Marc Stephenson IBM PSPA (Personal System Programming - Austin,TX) DISCLAIMER: The content of this posting is independent of official IBM position. marc@stingray.austin.ibm.com VNET: MARC at AUSVMQ IBM T/L: 793-3796
dcm@codesmith.austin.ibm.com (Craig Miller) (05/13/91)
In article <9357.282caa94@jetson.uh.edu> elee4fg@jetson.uh.edu writes: >From: seng Gan >Star Enterprise > >1) csh does not support suid. If your csh suid scripts file has this > #!/bin/csh > It won't work. > Your need to do this in ksh I believe this is not an AIXism but is a BSDism. The BSD4.3 csh source I have access to does not support suid either. This is documented, I believe. > /bin/csh <your suid script file> ( without #!/bin/csh ) Are you sure this really works? Suid scripts have never worked without "#!". Your script may run, but it won't run suid. >6) Increase your process limit > your need to edit /usr/include/sys/limits.h > CHILD_MAX to increase your number of process allowed per user id > in ksh 40 is too small for root > chdev -l sys0 -A maxuproc has no effect in ksh (a) /usr/include/sys/limits.h has no effect on process limits (unless you're rebuilding your kernel). Are you rebuilding your kernel? (b) I wasn't aware that max number of procs per user was enforced for root. If it is, it shouldn't be (IMHO). (c) What does ksh have to do with process limits? Can you give a more specific example of why "chdev -l sys0..." has no effect in ksh (vs sh or csh). Craig -- Craig Miller Internet: dcm@aixwiz.austin.ibm.com IBM Austin Vnet: tkg007 at ausvmq AIXV3 Change Team (level3) IBM internal: dcm@littleguy.austin.ibm.com "I do not represent IBM or any other respectable company."
Andreas.Kaiser@f7014.n244.z2.stgt.sub.org (Andreas Kaiser) (05/14/91)
>7) There is some strange limit to number of device attached to >SCSI card Basically, if you have internal device attached to SCSI >adapter you can only have one or two extenal device per adapter. It >applies in 540 and 530 also in addition to 320. I cannot imagine a reason for this, but it is obviously not an enforced limitation. We have both an internal 320 MB SCSI drive and 3 external SCSI devices (tape, CDROM, 1GB disk) running fine on a 320. Even with a forth external device, a mechanically identical Tandberg tape drive which did properly respond but was unusable due to IBM tape firmware peculiarities (such as the unusual ECC feature), the other SCSI devices ran without any trouble. >8) inetd >I let my rs/6000 PING EACH OTHER every hour. That seems to be >the only way to keep it running constantly. inetd running without any problems (connected to 6000, 6150, DOS-KA9Q). Gruss, Andreas * Origin: kaiser@ananke.stgt.sub.org - Stuttgart FRG (2:244/7014)
elee4fg@jetson.uh.edu (05/14/91)
In article <1991May13.115433.22669@bellcore.bellcore.com>, jona@iscp.Bellcore.COM (Jon Alperin) writes: > In article <9357.282caa94@jetson.uh.edu>, elee4fg@jetson.uh.edu writes: > |> From: seng Gan > |> Star Enterprise > |> > |> Here are the problems that I known about: > > < list trimmed > > |> 10) RS/6000 (token ring TCP/IP) connect to Cisco > |> Need to define static ARP route in Cisco to RS/6000 or else the link > |> is not stable. > > We have found that any workstations running YP/NIS need to have routes > defined statically, since AIX will set up these routes during a boot before getting the NIS information. Thus, if you use mnemonic names instead of addresses AIX cannot do an IP lookup against the host table (which it has not yet gotten from NIS), and t herefore the route is lost. > > |> Seng C, Gan > |> Unix SysAdm > |> Star Enterprise > |> Tel:(713)874-7816 > > -- > Jon Alperin > Bell Communications Research > > ---> Internet: jona@iscp.bellcore.com > ---> Voicenet: (908) 699-8674 > ---> UUNET: uunet!bcr!jona > > * All opinions and stupid questions are my own * I am not running YP/NIS but still need those static routes Seng C. gan
elee4fg@jetson.uh.edu (05/14/91)
In article <7513@awdprime.UUCP>, dcm@codesmith.austin.ibm.com (Craig Miller) writes: > In article <9357.282caa94@jetson.uh.edu> elee4fg@jetson.uh.edu writes: >>From: seng Gan >>Star Enterprise >> >>1) csh does not support suid. If your csh suid scripts file has this >> #!/bin/csh >> It won't work. >> Your need to do this in ksh > > I believe this is not an AIXism but is a BSDism. The BSD4.3 > csh source I have access to does not support suid either. This > is documented, I believe. > >> /bin/csh <your suid script file> ( without #!/bin/csh ) > > Are you sure this really works? Suid scripts have never worked > without "#!". Your script may run, but it won't run suid. > >>6) Increase your process limit >> your need to edit /usr/include/sys/limits.h >> CHILD_MAX to increase your number of process allowed per user id >> in ksh 40 is too small for root >> chdev -l sys0 -A maxuproc has no effect in ksh > > > (a) /usr/include/sys/limits.h has no effect on process limits (unless > you're rebuilding your kernel). Are you rebuilding your kernel? > > (b) I wasn't aware that max number of procs per user was enforced > for root. If it is, it shouldn't be (IMHO). > > (c) What does ksh have to do with process limits? Can you give a > more specific example of why "chdev -l sys0..." has no effect > in ksh (vs sh or csh). > > > Craig > -- > Craig Miller Internet: dcm@aixwiz.austin.ibm.com > IBM Austin Vnet: tkg007 at ausvmq > AIXV3 Change Team (level3) IBM internal: dcm@littleguy.austin.ibm.com > "I do not represent IBM or any other respectable company." a) It has no general process limits unless you are running ksh. Try it out. if you run ksh and you are root, the normal limit is 40 per user id. b) Yes it is enforced on root. c) That problems only occurs when you are in ksh. The process limits in ksh is defined in limits.h per user id. This problem is documented by IBM SystemXtra. Why don't people in IBM talk to each other??????? Seng C. Gan By the way, I run AIX 3.1 with 3002 patch on Model 540 and Model 530 Call me if you want to know the problem number. Tel:(713)874-7816
moody@snap.austin.ibm.com (05/16/91)
In article <674293502.1@ananke.stgt.sub.org> Andreas.Kaiser@f7014.n244.z2.stgt.sub.org (Andreas Kaiser) writes: > > >7) There is some strange limit to number of device attached to > >SCSI card Basically, if you have internal device attached to SCSI > >adapter you can only have one or two extenal device per adapter. It > >applies in 540 and 530 also in addition to 320. The limitation is due to the length of the cable necessary to make this happen. There is no other limitation. > >I cannot imagine a reason for this, but it is obviously not an enforced >limitation. True... > >8) inetd > >I let my rs/6000 PING EACH OTHER every hour. That seems to be > >the only way to keep it running constantly. Don't know about this stuff... > * Origin: kaiser@ananke.stgt.sub.org - Stuttgart FRG (2:244/7014) -- James Moody aixnet:moody@moody.austin.ibm.com Personal Systems Programming Austin VNET:MOODY@AUSVMQ AIX Field Support - Level 3 internet:moody@aixwiz.austin.ibm.com