Unsolved
This post is more than 5 years old
2 Intern
•
152 Posts
0
2352
July 20th, 2010 13:00
High disk read times for AIX database servers
We are experiencing high disk read times for 2 of the AIX database servers which are connected to the DMX-4. So, any one can tell me how to troubeshoot such issues. I checked on the DMX and everything looks normal.
0 events found
No Events found!
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
July 20th, 2010 14:00
do you have SPA, EMC Performance manager ? Have you tried using symstat for real time statistics and see what's going on with devices presented to those AIX boxes ?
ZepHead
88 Posts
1
July 20th, 2010 17:00
I would single out the device(s) first that you are seeing performance issues. What type of devices/configuration are these..? How are you determining this, another words - what tools are you using for determination..? Are you using PowerPath and if so, do you have 2 (or more) paths to a device as well as fully optimized...? What errors and/or symptoms are showing in "errpt"...? Are there other hosts experiencing the same issue..? What has changed lately..? Did it just start happening..? When did this start happening..? Do you have a performance PMR with IBM..? Any symptoms from a DBA perspective..?
vimad
2 Intern
•
152 Posts
0
July 21st, 2010 04:00
Its a 2 node AIX cluster and it seems the load is more on node 2 when compared to node 1 and at peak hrs node 2 is getting really hit hard. how can i get the stats on devices using symcli.. The paths are good. There are no other hosts experiencing the issues.
Quincy561
1.3K Posts
1
July 21st, 2010 04:00
You can use symstat to get statistics from the system, but a better choice might be to collect STP data with the storstpd deamon.
vimad
2 Intern
•
152 Posts
0
July 21st, 2010 06:00
Please let me know if this is the correct command to get the disk stats
symstat -type disk -i 5 -c 3 -sid abc -disk 0afb
where 0afb is a lun mapped and masked to a AIX server
Please correct me i am missing anything
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
1
July 21st, 2010 10:00
create a device group:
symdg create
add symdev devices presented to that box to a device group:
symld -g add dev XXXX -sid 123
symstat -g -i 10
this will give you symdev stats in 10 second increments.
vimad
2 Intern
•
152 Posts
0
July 21st, 2010 11:00
Dynamox thanks for the response.. The AIX cluster seems to be operating normal now. They had some perfromace issues and this errors were reported
SC_DISK_ERR4 and disk operation errors in the past. Any thoughts why this errors were generated. I was thinking this may be due to some path issues with switches and storage array
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
July 21st, 2010 20:00
i get a lot of hits on IBM web site ..worth opening a case with them so they can analyze the logs.