I have a database (IBM/Informix/Ardent UniVerse) that is running on a
Dell PowerEdge 6600 (stats below). Sometimes when there is a very high
I/O, any database command shell will hang when trying to exit. A ps
shows that the process is in a get_request_wait.
When the process that is causing the high I/O finishes what it is doing,
then about 5-10 seconds later the command shell that was trying to exit
will finish. I've read some threads on the list and elsewhere and seen
some patches that refer to "fixing" get_request_wait and
wait_on_page/wait_on_buffer "bugs."
Could mounting file systems onto directories that are only accessible
after another file system is mounted cause an issue (nested mount
points)? For example I have fs /dev/sdd1 that contains a directory
called 'dbms,' I then mount other filesystems onto directories contained
within dbms/:
/dev/sdd1 /cubs
/dev/sdc1 /cubs/dbms/ACCOUNT
/dev/sde1 /cubs/dbms/16kb
System specs are below. I would try 2.4.20 vanilla, but for some reason
I get a panic WRT some SMP things when 2.4.20 boots, but that's another
issue that I will try to tackle on my own.
Is there anything that I can try (patches, /proc settings, etc.) to help
alleviate this issue? Should I reconfigure my mount points? If more
information is needed, please let me know.
Thanks in advance,
Andy.
Andrew Rechenberg
Infrastructure Team, Sherman Financial Group
arechenberg@shermanfinancialgroup.com
------------------------------------------------------------------------
-------
Dell PE6600
(4) Xeon 1.4GHz with HT enabled
8GB RAM
Red Hat 7.3
2.4.18-19-7.xbigmem with SD_TIMEOUT in sd.c upped to 90*HZ and
megaraid2.[c,h]
(2) Dell PERC3/QC (megaraid2 2.00.2) with most recent firmware
(1.74:3.27]
(1) Dell PERC3/DC (megaraid2 2.00.2)
[root@box ~]# ps -eo pid,stat,pcpu,nwchan,wchan=WIDE-WCHAN-COLUMN -o
args | grep wait
20 DW 0.1 19e8a6 get_request_wait [bdflush]
21 DW 0.3 19e8a6 get_request_wait [kupdated]
1440 S 0.0 1217c8 wait4 -bash
1580 S 0.0 1217c8 wait4 su -
1581 S 0.0 1217c8 wait4 -bash
1748 S 0.0 1217c8 wait4 -bash
1850 S 0.0 1217c8 wait4 /bin/bash
1918 S 0.0 1217c8 wait4 su -
1919 S 0.0 1217c8 wait4 -bash
2158 S 0.0 1217c8 wait4 /bin/bash
2430 S 0.0 1217c8 wait4 -bash
2611 S 0.0 1217c8 wait4 su -
2612 S 0.0 1217c8 wait4 -bash
2869 S 0.0 1217c8 wait4 -bash
3013 S 0.0 1217c8 wait4 su -
3014 S 0.0 1217c8 wait4 -bash
3425 S 0.0 1217c8 wait4 su -
3427 S 0.0 1217c8 wait4 -bash
3534 S 0.0 1217c8 wait4 -bash
3673 S 0.0 1217c8 wait4 su -
3674 S 0.0 1217c8 wait4 -bash
3753 S 0.0 1217c8 wait4 uv
3853 D 6.7 19e8a6 get_request_wait RESIZE -N ALECREDCBI1 18 29826
128
3882 S 0.0 1217c8 wait4 uv
3913 D 0.0 19e8a6 get_request_wait uv
3929 D 4.4 19e8a6 get_request_wait RESIZE -N PFMODDEBTOR3 2 342313
8
3939 D 0.2 19e8a6 get_request_wait uv
3954 D 0.1 19e8a6 get_request_wait uv
3956 S 0.0 14d79c pipe_wait grep wait
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Jan 23 2003 - 22:00:25 EST