System hang due to high iowait

[csu4 at fedex.com]

Wayne,

Looks like it does not work. Now I have the same problem even
with ext2 filer system. My box is not running e-mail server, but
as a file server. The data transfer volume is huge (70GB+), but
the individual file is not that big, 200MB in average. After the
system hang, I could see the rcp (which is the program I transfer
files) is still running, just because the iowait too high causing
the server is not moving at all. The server still can be pinged,
BTW.




Thx & Rgds,


Chang Hong








                                                                                                                 
                    "Wayne Pinette"                                                                              
                    <Wpinette at cariboo.bc       To:     <redhat-list at redhat.com>                                  
                    .ca>                       cc:                                                               
                    Sent by:                   Subject:     Re: System hang due to high iowait                   
                    redhat-list-bounces@                                                                         
                    redhat.com                                                                                   
                                                                                                                 
                                                                                                                 
                    2004-10-07 23:24                                                                             
                    Please respond to                                                                            
                    General Red Hat                                                                              
                    Linux discussion                                                                             
                    list                                                                                         
                                                                                                                 




Hmmm Im not sure.  All of my drives are ext3 drives (including the one
with that patch I sent).
I know the server which I had to do it on isn't handling large data
files, but rather many many many
small ones (mail server with anti-virus scan which handles ~1800
messages an hour so it's all disk io).

Im still guessing you want to mess with the pagecache settings.
Perhaps try a larger number.
(2 10 75 /proc/sys/vm/pagecache) .


Wayner





>>> csu4 at fedex.com 06/10/2004 11:32:52 pm >>>

Hi Wayne,

Thanks a lot for the tips. I did have a try:
1. If the file system is ext2 for the large IO operation,
   then your work around works;
2. If the file system is ext3, then it hangs again.

I'm wondering if this is ext3 problem or the parameter you
mentioned must be adjusted for ext3?





Thx & Rgds,


schyu










                    Wayne Pinette

                    <Wpinette at cariboo.bc       To:
redhat-list at redhat.com
                    .ca>                       cc:

                    Sent by:                   Subject:     Re: System
hang due to high iowait
                    redhat-list-bounces@

                    redhat.com





                    2004-10-04 23:19

                    Please respond to

                    General Red Hat

                    Linux discussion

                    list







I had a similar problem and did a google search on it a while back.  I
found this little snippet and
tried it and have never had a problem since :

In your rc.d/rc.local file append the following two lines :


echo 100 > /proc/sys/vm/inactive_clean_percent
echo 2 10 20  > /proc/sys/vm/pagecache


I would give this a try.

Wayner


>>> csu4 at fedex.com 04/10/2004 1:46:04 am >>>
Hi there,

Could anyone help to find the system hung problem? The box
is running RHEL 3AS Update3. Whenever I do a massive data
transfer (such as rcp from another box to this box), after
certain time (around six to seven hours), this box would
hang up. I could see the top output even it's hung, and it
showed that the iowait is 99% and system contributed another
1% making the server moving nowhere.

The box connected to a HP MSA500 Smart Array Storage, thru
multipath (I enabled it following the user guide) connection.
The filesystem is ext3 format. Any problem here?





Thx & Rgds,


SCHYU







--
redhat-list mailing list
unsubscribe mailto:redhat-list-request at redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list

--
redhat-list mailing list
unsubscribe mailto:redhat-list-request at redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list



--
redhat-list mailing list
unsubscribe mailto:redhat-list-request at redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list

--
redhat-list mailing list
unsubscribe mailto:redhat-list-request at redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/redhat-list