SuSE SLES 8: raw devices broken after kernel update?

SuSE SLES 8: raw devices broken after kernel update?

Post by Matthias Krauss » Tue, 06 May 2003 05:45:17



Hi,

anybody out there running IDS on a SuSE Linux Enterprise Server 8
("SLES 8" aka United Linux 1.0 for x86) with all recommended
kernel/glibc updates installed (preferably on a Dual-Xeon box or
similar)?

We wanted to test if our application ("classic" 4GL) works with the
brand new IDS 9.40 on linux. After installing, we encountered the
following problem:
Our box locks up (freezes) completely after "oninit -iv" if the rootdbs
chunk is a raw device. Can be reproduced both with IDS 7.31.UD2 and with
9.40.UC1-1. No /var/log/messages entry, no kernel oopses, no assertion
failure - absolutely nothing to deal with. The first command typed in
after the "oninit ..." will not return (even simple ones like "ls").
The only solution is to press the reset button.
If the chunk is a flat file, both (7.31/9.40) initialize fine.

I did some testing, proceeding from a vanilla SLES 8 installation
straight from the CD-ROM (that is, without any online updates) and
then installing all SuSE-recommended critical (read: kernel + glibc)
patches in chronological order via YOU ("YaST Online Update";
seems to be similar to RH's "up2date").
Here are the results:

STEP 1: stock SLES 8 install without any online updates
        7.x initializes w/o problems
        9.40 gives up after "Bringing up ADM VP ...", leaving some
        shm segments (had to remove them via ipcrm -m ...)
        This was discussed in another thread some days ago;
        there is a glibc update for SLES 8 that fixes this problem;
        it is similar to the patch Marion Kussujew suggested for
        SuSE Professional/Personal 8.1 (see STEP 2)

STEP 2: after installing the (recommended from SuSE) glibc update
        both (7.31/9.40) initialize correctly and run fine on
        flat files and on raw devices

STEP 3: after installing the first kernel security update
        (released 15. Nov 2002)
        nothing's changed, both servers run fine

STEP 4: after installing the (SuSE recommended) kernel update "patch-4922"
        (released 21. Feb 2003)
        both servers lock up the whole box after "oninit -iv"

SuSE's short description of "patch-4922" is:
 This kernel contains several bugfixes, performance
 improvements and driver updates. Use of this kernel is
 mandatory when you plan to run Oracle on UnitedLinux 1.0
 based systems and recommended in all other cases.      

Isn't that ironic? IDS not running after a kernel update needed for
Oracle certification.

I'd be happy if anybody could shed some light on this or share some
opinion running IDS 7.x/9.x on SuSE's SLES 8.

At the moment I really don't know how to sort things out.
IBM Informix states that they are certifying linux based on
glibc/kernel version.
According to this, both servers should work on SLES 8.

Unfortunately,
- after installing the latest kernel security patch from SuSE
  for SLES 8 (ptrace-bug related) I cannot even strace the oninit process
- SuSE's patches are (at least the kernel-related ones,IIRC) cumulative;
  if you omit one of them, you'll lose the "whole chain"
- we have to return the box next Tuesday for some other reason, and
  we have no other Dual-CPU machine for the moment. So I won't be able
  to do some in-depth testing.  

Sidenote: It makes no difference whether
- we enable/disable Hyperthreading in BIOS
- we enable/disable NOAGE,RESIDENT and/or MULTIPROCESSOR in $ONCONFIG
- we boot with/without some safe kernel parameters (acpi=off disableapic
  ...)
- the raw devices ar bound to a LVM's logical volume or directly to
  a disk partition like /dev/sda6

--
Matthias Krausse
PAS GmbH Herdecke
58313 Herdecke
Germany

 
 
 

SuSE SLES 8: raw devices broken after kernel update?

Post by eu.. » Sun, 25 May 2003 01:20:33


Hi,

I got a similar intialization problem with IDS 9.40.UC1 on SCO linux
4.0 (based on United Linux 1.0, so very near to SuSE linux 8).
With a default initialization (no raw devices, default low values for
buffers, lock, shmvirtsize...), the oninit -iv command stops at


Checking group membership to determine server run modesucceeded
Reading configuration file
'/usr/informix/etc/onconfig.oltlc7'...succeeded
Creating /INFORMIXTMP/.infxdirs ... succeeded
Creating infos file "/usr/informix/etc/.infos.oltlc7" ...
"/usr/informix/etc/.co
nf.oltlc7" ... succeeded
Writing to infos file ... succeeded
Checking config parameters...succeeded
Allocating and attaching to shared memory...succeeded
Creating resident pool 1382 kbytes...succeeded
Creating buffer pool 10002 kbytes...succeeded
Initializing rhead structure...succeeded
Initializing ASF ...succeeded
Initializing Dictionary Cache and SPL Routine Cache...succeeded

On the same machine IDS 9.30 works fine...
It's smell as something like a bug (??).

I don't know.


> Hi,

> anybody out there running IDS on a SuSE Linux Enterprise Server 8
> ("SLES 8" aka United Linux 1.0 for x86) with all recommended
> kernel/glibc updates installed (preferably on a Dual-Xeon box or
> similar)?

> We wanted to test if our application ("classic" 4GL) works with the
> brand new IDS 9.40 on linux. After installing, we encountered the
> following problem:
> Our box locks up (freezes) completely after "oninit -iv" if the rootdbs
> chunk is a raw device. Can be reproduced both with IDS 7.31.UD2 and with
> 9.40.UC1-1. No /var/log/messages entry, no kernel oopses, no assertion
> failure - absolutely nothing to deal with. The first command typed in
> after the "oninit ..." will not return (even simple ones like "ls").
> The only solution is to press the reset button.
> If the chunk is a flat file, both (7.31/9.40) initialize fine.

> I did some testing, proceeding from a vanilla SLES 8 installation
> straight from the CD-ROM (that is, without any online updates) and
> then installing all SuSE-recommended critical (read: kernel + glibc)
> patches in chronological order via YOU ("YaST Online Update";
> seems to be similar to RH's "up2date").
> Here are the results:

> STEP 1: stock SLES 8 install without any online updates
>         7.x initializes w/o problems
>         9.40 gives up after "Bringing up ADM VP ...", leaving some
>         shm segments (had to remove them via ipcrm -m ...)
>         This was discussed in another thread some days ago;
>         there is a glibc update for SLES 8 that fixes this problem;
>         it is similar to the patch Marion Kussujew suggested for
>         SuSE Professional/Personal 8.1 (see STEP 2)

> STEP 2: after installing the (recommended from SuSE) glibc update
>         both (7.31/9.40) initialize correctly and run fine on
>         flat files and on raw devices

> STEP 3: after installing the first kernel security update
>         (released 15. Nov 2002)
>         nothing's changed, both servers run fine

> STEP 4: after installing the (SuSE recommended) kernel update "patch-4922"
>         (released 21. Feb 2003)
>         both servers lock up the whole box after "oninit -iv"

> SuSE's short description of "patch-4922" is:
>  This kernel contains several bugfixes, performance
>  improvements and driver updates. Use of this kernel is
>  mandatory when you plan to run Oracle on UnitedLinux 1.0
>  based systems and recommended in all other cases.      

> Isn't that ironic? IDS not running after a kernel update needed for
> Oracle certification.

> I'd be happy if anybody could shed some light on this or share some
> opinion running IDS 7.x/9.x on SuSE's SLES 8.

> At the moment I really don't know how to sort things out.
> IBM Informix states that they are certifying linux based on
> glibc/kernel version.
> According to this, both servers should work on SLES 8.

> Unfortunately,
> - after installing the latest kernel security patch from SuSE
>   for SLES 8 (ptrace-bug related) I cannot even strace the oninit process
> - SuSE's patches are (at least the kernel-related ones,IIRC) cumulative;
>   if you omit one of them, you'll lose the "whole chain"
> - we have to return the box next Tuesday for some other reason, and
>   we have no other Dual-CPU machine for the moment. So I won't be able
>   to do some in-depth testing.  

> Sidenote: It makes no difference whether
> - we enable/disable Hyperthreading in BIOS
> - we enable/disable NOAGE,RESIDENT and/or MULTIPROCESSOR in $ONCONFIG
> - we boot with/without some safe kernel parameters (acpi=off disableapic
>   ...)
> - the raw devices ar bound to a LVM's logical volume or directly to
>   a disk partition like /dev/sda6


 
 
 

SuSE SLES 8: raw devices broken after kernel update?

Post by Walter Haslbec » Mon, 26 May 2003 19:37:04


Hello,

[...]

Same happens here with SuSE Linux 8.2

Any hints?

Walter

 
 
 

SuSE SLES 8: raw devices broken after kernel update?

Post by Richard Kofle » Wed, 28 May 2003 05:58:10



> Hello,


> [...]


> Same happens here with SuSE Linux 8.2

> Any hints?

> Walter

google for my posting from 16th May 2003
speaking about the necessary glibc update
available from SuSE.
This info & the URL poiting to
the right download I got from M. Fuerderer, IBM

HTH
dic_k

 
 
 

SuSE SLES 8: raw devices broken after kernel update?

Post by Walter Haslbec » Wed, 28 May 2003 15:06:53



> google for my posting from 16th May 2003
> speaking about the necessary glibc update
> available from SuSE.

you wrote about a glibc update for SuSE Linux 8.1. But there is no glibc
update (yet) for SuSE Linux 8.2 (which contains glibc-2.3.2-6).

All software sucks.

Walter