RE: HPL hangs server
> -----Original Message-----
> From: Pravin Kedia [SMTP:kedia@us.ibm.com]
> Sent: Wednesday, August 27, 2003 10:23 AM
> To: Bill Dare
> Cc: 'informix-list@iiug.org'; owner-informix-list@iiug.org
> Subject: Re: HPL hangs server
>
>
>
>
>
>
> Hi Bill,
>
> Allocating lot of shared memory segment is normal.
> It depends on the load that you are trying to do.
>
> Few questions:-
> 1. Are u using multiprocessor machines? (How many processors)
[Bill Dare]
10
> Try following things:-
>
> 1. setting PLOAD_SHMAT=1 and running these multiple onpload.
[Bill Dare]
I'm running the loads right now with PLOAD_SHMAT set. No failure
yet. Can't get onstat -j to work.
> 2. When the server hangs, can you capture the following,(during the loads)
>
> onstat -g ath
> onstat -g stk all>
> onstat -j <path and the pload log file name>
[Bill Dare] I don't find any documentation on onstat -j except the usage message
which isn't much help. What am I doing wrong?
proddb:[/usr/informix/work]> onstat -j /tmp/logfile
onpload attach failed, rtn: -1 errno: 0
shmkey: ABCDEF load_base: 0
proddb:[/usr/informix/work]>
I do have 3 onploads running at any point in time. How does onstat
know which to connect to?
> > seg
> > ath
> > stk all
>
> Thanks,
> Pravin.
>
>
>
>
>
> Bill Dare
>
> <dareb@jevic.com> To:
> "'informix-list@iiug.org'"
> Sent by: <informix-list@iiug.org>
>
> owner-informix-li cc:
>
> st@iiug.org Subject: HPL hangs server
>
>
>
>
>
> 08/26/2003 03:52
>
> PM
>
>
>
>
>
>
>
>
>
> 9.40.FC2
> AIX 5.2
>
>
> I am having a problem with HPL causing the server to hang. I have a case
> open with tech support (374620) but we are getting no where. Anybody run
> into this before. I am running multiple onploads from a script to load a
> new database. It runs for a while, sometimes 10-15 minutes, sometimes
> longer. The server hanging up always seems to be a little different, but
> most times I have to kill -9 the master oninit daemon. No assertion
> failure. The failure occurs more often if I use a TCP/IP connection and
> even more often if I run multiple onploads instead of just one. I always
> get the following error from onpload:
>
>
> HPL failpoint: 11014
>
>
> 16:36:33 pl_main.c, line 569, thread -1, proc id 741404, Segmentation
> Violation.
> 16:36:33 PANIC: Attempting to bring system down>
>
>
>
>
> Also, each time a new onpload starts I get:
>
>
> Tue Aug 26 16:26:02 2003
>
>
> 16:26:02 Dynamically allocated new virtual shared memory segment (size
> 4096KB)>
>
>
>
>
> Is this normal. I've got plenty of virtual shared memory, 4GB.
>
>
> Any ideas.
>
>
> Regards,
> Bill
>
>
>
>
>
sending to informix-list